Date post: | 25-Nov-2014 |
Category: |
Documents |
Upload: | sprtoshbti |
View: | 159 times |
Download: | 0 times |
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0101-1
TEMPORAL DIFFERENCES IN PROCESSING OF SYMMETRIC OBJECTS IN HUMAN
BRAIN
1Taslima Ahmed,
2 Umesh Kumar,
3Prasant kumar
1,2&3
IIMT College of Engineering, Greater Noida, UP
ABSTRACT
Despite ample of research, perception, recognition and
identification of symmetric objects like faces are still a
matter of active debate. Here we studied the difference in the
neural mechanisms manifested by the most symmetrical
natural objects, “human faces” and unnatural symmetrical
objects like “paired ears and non biological objects” by
using event related potential (ERP) study. To test such an
assumption, human faces and symmetric unnatural objects
were presented to subjects on a projected screen as visual
stimuli. Electrophysiological data showed an important
latency effect for both of these categories of objects. N200
represents an important step in recognition of symmetric
objects in visual world. Moreover the amplitudes of
unnatural symmetrical objects are larger compared to those
for facial activation, which confirms that the neural
activation is more dominant for the unnatural novel objects
than natural objects.
Index Terms— Event related potentials (ERPs); N200;
Electrophysiology; Latency.
1. INTRODUCTION
Advances in electrophysiological techniques and brain
imaging technology (especially functional magnetic
resonance Imaging (fMRI)) have radically improved our
understanding of the functional organization, spatial
locations as well as the process underlying the understanding
of spatial world around us. The burgeoning literature on this
topic of spatial localization indicates that object recognition
varies widely across brain system in visual cortex giving rise
to highly specific altercation as a domain-specific or
domain-general. The organization of the ventral visual
pathways is characterized by strong selectivity for particular
object categories at the level of both individual neurons and
larger cortical regions [1]. There is a region in the brain,
lateral occipital cortex (LOC) which is known as the object
selective region [2]. The LOC has little selectivity for
particular stimulus categories [3,4], but several regions of
cortex near the LOC are selective for particular object
categories; they respond at
least twice as strongly to their preferred stimuli than no other
stimuli. All human cortical regions can be found that
respond selectively to faces (the fusiform face area (FFA)
[5,6] and, in many individuals, the occipital face area (OFA)
[7,8], to places and to a lesser extent – to buildings (the
parahippocampal place area (PPA) [9,10]), to body parts
(the extrastriate body area (EBA)[11] and, in most people,
the fusiform body area [12] and to visually presented words
or letter strings [13].The location and functional properties
are very similar across humans. Hemodynamic techniques
such as fMRI have identified brain regions associated with
face-processing, most notably in fusiform, ventral occipital
and superior temporal cortex. The precise role of the lateral
midfusiform (LMF) with processing the invariant aspects of
faces is contentious however. Furthermore, though it is
generally activated during face perception, when comparing
face versus non-face stimuli [6], it is not always activated
during face recognition, when compromising familiar versus
unfamiliar faces. For reasons not fully understood however,
other studies have found effects of both face recognition and
priming in the LMF.
The temporal characteristics of face-processing have
been elucidated by electrophysiological techniques.
Intracranial ERPs reveal a negative potential peaking -
200ms post stimuli, in both ventral and lateral temporal
regions, which is greater to faces than scrambled faces or
non-faces objects [14]. These data suggest that
hemodynamic correlates of recognition of face priming the
fusiform area of occipital cortex of human brain, which is
mostly domain-specific.
The studies reviewed above have not directly compared
the process of similar object perception, recognition in a
particular localized area (domain specific) of the human
brain. Therefore, in the present study, we used even related
potentials study within a common paradigm. This allowed us
to compare these processes in terms of their cerebral
localization. We operational zed domain-general theory of
neural perception by comparing the ERPs study of faces vs.
unnatural objects like paired ears as described in later
sections.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0101-2
2. MATERIALS AND METHODS
2.1. Participants
Ten volunteers participated in the said ERPs study after
giving written and informed consent. Participants were six
men and four women, aged 19-30years (median 24). All
volunteers were right handed, in good health and with no
known neurological disorder.
2.2. Experimental design approach
We used 32-channel EEG system to record neuronal activity
during presentations of two categories of visual stimuli:
faces and symmetric pairs of biological objects (Fig. 1). The
photographs of faces were from an image database that has
been previously used in the lab. The pairs of ears were put
together from pictures of people‟s left and right ears. Just
like the left and right sides of a face, a person‟s left and right
ears are similar but are not exactly the same. The parameters
for the signal acquisition were used to acquire the ERPs of
100 objects from each category, stimuli onset time was
800ms while the inter-stimuli interval was fixed to 1s. By
using advanced soft computing techniques, brain regions
that are activated by faces and symmetric unnatural objects
were localized as regions of interests (ROIs).
Fig 1: Samples of photos of faces and symmetric unnatural
objects used as stimuli
2.3. Event related potentials (ERPs)
Event-related potentials (ERPs) in EEGs directly measure
the electrical response of the cortex to sensory, affective or
cognitive events. They are the voltage fluctuations in the
EEG induced within the brain, which are typically generated
in response to peripheral or external stimulations, and
appear as somatosensory, visual, and auditory brain
potentials, or as slowly evolving brain activity observed
before voluntary movements or during anticipation of
conditional stimulation [14]. The amplitude of ERPs ranges
from 1-30µV, lying relatively low in the background EEG
activity. Therefore, often there is a need for the use of signal
averaging procedure for their elucidation. The ERP
waveform can be quantitatively characterized across three
main dimensions: amplitude, latency, and scalp distribution.
The ERP signals are either positive such as P300 (positive
peak appearing 300ms after stimulus onset), or negative
such as N170 and N100 (negative peak appearing
170/100ms after stimulus onset). The digits indicate the time
in terms of milliseconds after the sensory stimuli (audio,
visual, or somatosensory).
Although determination of the location of the ERP
sources within the brain is a difficult task, the scalp
distribution of an ERP component can often provide very
useful and complementary information to that derived from
amplitude and latency. The scalp recorded ERP voltage
activity reflects the summation of both cortical and sub
cortical neural activity within each time window. On the
current source density (CSD) maps reflect primary cortical
surface activity. CSD maps are useful for forming
hypotheses about neural sources within the superficial
cortex.
2.4. Data processing
Analysis of electro encephalogram (EEG) data was
performed with MATLAB R2007a. The signal
corresponding to each visual stimulus was cut for averaging
purpose to get the event related component. Pre-stimulus
data of duration 200ms was considered for each epoch to
eliminate base line drift. Therefore, the total length of each
epoch is 200ms (pre-stimuli) and 800ms (stimuli onset),
which is shown in the figures.
2.4.1. ERP analysis
Following artifact removal, the EEG signal was down-
sampled to 256Hz. The time locked data were then filtered
with a 0.5-50Hz band pass filter. Epochs of -200ms before
stimulus onset and 800ms after stimulus onset were averaged
separately for faces and symmetrical objects.
3. RESULTS AND DISCUSSION
In the specific context of symmetric image-processing, an
ERP component is typically observed in the sensors at the
occipito-temporal region and referred to as the N200. It is
characterized by negativity about 200 milliseconds
following the onset of symmetric-images. The amplitude of
the N200 is significantly reduced for face images, as shown
in the figure below. A large body of work has supported the
idea that the N170 appears for a stimulus similar to a face in
appearance. However, the N200 appears to be sensitive for
symmetric objects. The N200 amplitude and latency were
measured at the maximum negative peak of the ERP for
faces and symmetric objects within 170-210ms.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0101-3
To assess the pattern of the N200 response for faces and
symmetric objects in parietal, temporal and occipital
electrodes (T5, O1, P3, O2, T6, P4), two ways ANOVA
with hemisphere and category (face, symmetric objects) as
repeated measures and amplitude or latency as dependent
measures were selected. Analysis reveals no difference
between the pattern of responses to faces and symmetric
objects at the right and left occipito-temporal electrodes.
The amplitude of the N200 was however larger for
symmetric novel objects than faces [F(1,9)=37.7, p<0.001].
The figures (3, 4) show the variation of amplitudes in
various electrodes.
The latency of the N200 for novel objects and faces
were very similar. A one-way ANOVA with hemisphere as
repeated measures revealed no difference in latency between
the components during two different objects condition
[F(1,9)=2.3, p=0.17].
The results of this study unveiled a number of
dissociations between different forms of neural activity
associated with early face and object processing. The
amplitude of the N200 was higher in response to non-face
components than in response to face components. This result
links the N200 to a symmetric object detection mechanism
with the facial recognition mechanism.
Fig 2: 10-20 system of EEG electrode placement
Fig 3: ERP component from left occipital (O1) region for
faces and unnatural symmetric objects
Fig 4: ERP component from right occipital (O2) region for
faces and unnatural symmetric objects
4. CONCLUSION
By using novel techniques, brain regions that were activated
by natural and unnatural symmetric objects were
distinguished. Consistent with previous studies, fusiform
gyrus (especially right and left FFA, O1 & O2) was the most
reliable region that was activated by faces and novel objects,
though the amplitudes of activation for faces were smaller
compared to objects. The results also provided some clue as
to the „domain general concept‟ of the human brain with
respect to similar types of visual inputs in „domain specific
regions‟. The activation level indicates the neuronal activity
during the recognition of novel objects, which are peculiar
in nature; hence the amplitudes of activation in such cases
are higher in comparison to faces, which are important and
common images recognized by the human visual system.
REFERENCES
[1] H.P. Beeck, J. Haushofer and N.G. Kanwisher, “Interpreting
fMRI data: maps, modules and dimensions,” Reviews, J.
Neuroscience,Nature, vol.9, pp. 123-135, 2008.
[2] R.Malach, et al., “Object-related activity revealed by functional
magnetic resonance imaging in human occipital cortex,”. Proc.
Natl. Acad. Sci. USA, vol.92, pp. 8135-8139, 1995.
[3] K.G. Spector, Z. Kourtzi and N. Kanwisher, “The lateral
occipital complex and it‟s role in object recognition,” Vision Res.,
vol.41, pp. 1409-1422, 2001.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0101-4
[4] N.Kanwisher, R.P. Woods, M. Lacoboni and J.C. Mazziotta,
“A locus in human extrastriate cortex for visual shape analysis,”
J.Cogn. Neurosci., vol.9, pp. 133-142, 1997.
[5] N.Kanwisher and G. Yovel, “The fusiform face area: a cortical
region specialized for the perception of faces,” Philos. Trans. R.
Soc. Lond. B Biol.Sci., v. 361, pp. 2109-2128, 2006.
[6] N.Kanwisher, J. McDermott and M.M. Chun, “The fusiform
face area: a module in human extrastriate cortex specialized for
face perception,” J.Neurosci., vol.17, pp. 4302-4311, 1997.
[7] K.G. Spector, “The neural basis of object perception,”
Curr.Opin. Neurobiol., vol. 13, pp. 159-166, 2003.
[8] B.Rossion et al., “A network of occipito-temporal face
sensitive areas besides the right middle fusiform gyrus is necessary
for normal face processing,”Brain, vol. 126, pp. 2381-2395, 2003.
[9] R. Epstein and N. Kanwisher, “A cortical representation of the
local visual environment,” Nature, vol. 392, pp. 598-601, 1998.
[10] M.V. Peelen and P.E. Downing, “The neural basis of visual
body perception,” Nature Rev. Neurosci., vol. 8, pp. 636-648,
2007.
[11] R.F. Schwarzlose, C.I. Baker and ] N.Kanwisher, “Separate
face and body selectivity on the fusiform gyrus,” J.Neurosci.,vol.
25, pp. 11055-11059, 2005.
[12] B.D.McCandliss,L.Cohen and S. Dehaene, “The visual word
form area: expertise for reading in the fusiform gyrus,” Trends
Cogn.Sci., vol.7, pp. 293-299, 2003.
[13] T.Allison, A. Puce, D.D. Spencer, G.McCarthy,
“Electrophysiological studies of human face perception: potentials
generated in occipitotemporal cortex by face and non-face stimuli,”
vol. 9, pp. 415-430, 1999.
[14] B. Blankertz, K.R. Muller, G. Curio and T.M. Vaughan, “The
BCI competition:progress and perspective in detection and
discrimination of EEG single trials,”IEEE Trans. Biomed. Engg.
Vol. 51, pp. 1044-1051, 2003.
CONFERENCE ON ―SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)‖ MARCH 26-27 2011
COS0102-1
Modelling Neuron for Biomedical Applications: A Review
1Taslima Ahmed , 2 Dr Jiten Ch Dutta
1Dept. of Electronics and Instrumentation Engg. IIMT College of Engineering, Greater Noida, Knowledge Park-III, UP-201306, India
2Dept of ECE, Tezpur University , Napaam Post, Tezpur, Assam – 784 028, India
Abstract: Modelling of neuron including the action of synapse has played an important role in the
field of biomedical engineering and neurology for simulation of receptor function and electrical
activity of the postsynaptic neuron. In this paper, we review some literatures concerning the
development of different neuron models giving special emphasis on Dutta and Roy models.
Keywords— Neuron, Synapse, MOSFET, Postsynaptic membrane.
Over past few years, many electronic circuits
have been developed to reproduce the behaviour of
nerve axons [1]-[5]. A very good account of this
type of modelling is reviewed by Harmon et al [6]
and Lewis [7]. But among these models,
neuroscientists have, so far, utilized Hodgkin-
Huxley (H-H) model as a circuit analog of the
axonal membrane. In this model, the capacitance of
the lipid bilayer of postsynaptic membrane is
represented by CM and is found to be constant and
the membrane resistance is determined in terms of
three parallel conductances gNa, gk, and g0 as shown
in Fig.1. The conductances gNa, gK, and go represent
the membrane permeability of Sodium, Potassium
and other ions respectively. ENa, and EK are
respectively the chemical potentials of Sodium and
Potassium i.e., Nernstian membrane potential for
Sodium and Potassium. EO is the resting potential.
The gk and gNa conductances are found to be time
and voltage dependent.
Fig.1: H-H model
The total current in this model is given by:
I = Im+Io-INa+IK (1)
If Vm be the postsynaptic membrane potential
established by the ionic and capacitive membrane
current then
I = C(dVm/dt)+gO(Vm–EO)–gNa(Vm–ENa)+gK
(Vm–EK) (2)
The equations (1) and (2) are called H-H equations
which are simple and capable of explaining the
activity of neuron with the help of variable
permeability of membrane for different ions, e.g.,
sodium, potassium and other ions. But this model
has not explained the function of synapses on which
the variable permeability of postsynaptic membrane
arises.Refering to the biological activities of neuron,
the primary mode of communication between two
neurons is a biochemical process that occurs at
synapse. Synapse is essentially a junction called
synaptic cleft between two neurons namely
presynaptic and postsynaptic neurons. Signal from
presynaptic neuron to postsynaptic neuron is
transmitted through neurotransmitters released by
presynaptic neuron terminals in to the synaptic cleft.
Neurotransmitters diffuse through the cleft and then
bind with the specific receptor sites of the
membrane of postsynaptic neuron. This binding
mechanism initiates the opening of transmitter
CONFERENCE ON ―SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)‖ MARCH 26-27 2011
COS0102-2
gated ion channels resulting in to flow of ions into
the cell or out of the post synaptic cell.
The membrane of post synaptic neuron has two
types of ion channels – excitatory and inhibitory.
The excitatory channels are those which are specific
to sodium ions and inhibitory channels are those
which are specific to Chloride ions. The flow of
Sodium ions into the cell causes a membrane
potential called excitatory postsynaptic membrane
potential (EPSP) whereas the flow of Chloride ions
causes an inhibitory postsynaptic membrane
potential (IPSP).The electrical mechanism of
synapse is shown in Fig 2. When an action potential
from the presynaptic neuron arrives at its terminals
connecting the cleft, neurotransmitters are released
into the cleft which diffuse through the cleft and
bind with the receptor sites of the postsynaptic
membrane. This binding mechanism opens the ion
channels situated at the membrane surface and ions
move into or out of the membrane. If the synapse is
excitatory, Sodium ions flow into the cell resulting
into positive current. As a result the membrane
depolarizes. If sufficient number of Sodium
channels open, then membrane potential will be
greater than the threshold VT of the neuron and
initiates an action potential. If the synapse is
inhibitory, Chloride ions move into the cell,
resulting into negative current. As a result the
membrane hyperpolarizes. If the numbers of
opening of Chloride channels are sufficiently large
then membrane potential will be able to initiate an
action potential in negative direction. The
presynaptic equivalent circuit is shown in Fig 3(a),
where I is the total current from ionic channels of
all synapses and E1 ,E2, …., EM represent the
chemical potentials of each corresponding ions. For
example, EM may be ENa or may be ECl. The total
current I will stimulate the postsynaptic neuron to
initiate an action potential [8]. Fig 3(b) shows the
equivalent circuit of a synapse which is developed
by adding H-H equivalent circuit with the
presynaptic circuit shown in Fig 3(a) [9].The
postsynaptic membrane consists of a lipid bilayer
and transmembrane protein ion channels. Some ion
channels such as sodium, chloride etc. are
controlled by the neurotransmitters that bind with
the receptor sites, i.e. the
Fig.2: Electrical mechanism of synapse
Fig.3(a): Equivalent circuit of a presynaptic neuron
Fig. 3(b): Electrical equivalent circuit of synapse
amount of ionic current is dependent upon the
activity of the transmitter-receptor binding. In
simplest case, the binding reaction may be
represented as
CONFERENCE ON ―SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)‖ MARCH 26-27 2011
COS0102-3
Neuro-transmitter+Receptor(Closed)1
2
K
K
Neuro-
transmitter –Receptor(Open) (3)
Where K1 and K2 are the forward and backward
rate constants respectively. The transmitter gated
channels, therefore, have variable conductance
dependence on the binding activity of transmitters.
Dutta and Roy therefore have modeled transmitter
gated ion channels by MOSFET, because MOSFET
functions as a voltage controlled conductance in its
linear region [10]. In this model they have
considered gate voltage as a time dependent voltage
given by
Vg(t)=V0[(1–exp(-k1t)+exp(-k2t)U(t-tm)]
(4)
Where K1 and K2 are time constants analogous to
the rate constants of equation (3), U(t-tm) is the
Heaviside function andVo is a voltage proportional
to the maximum attainable conductance, when all
the transmitter-gated channels for a specific ion are
open. Based on this, they have developed a
biologically motivated model as shown in Fig.4 (a).
Their circuit models both for excitatory and
inhibitory synapses are shown in Fig.4(b) and 4(c)
respectively. In both these models they have
divided the postsynaptic membrane into three
patches to represent spatial summation of the
sodium current and chlorine current controlled by
respectively sodium and chloride conductances [10].
Fig.4(a): Biologically motivated model of
postsynaptic membrane.
Fig.4(b): Circuit model for excitatory action of
synapse
Fig.4(c): Circuit model for inhibitory action of
synapse
The simulation results from this model [Fig.5]
indicate that this model can be used in neuro
bioengineering area for simulation of
neurotransmitter-receptor binding activity and
electrical activity of the postsynaptic neuron.
CONFERENCE ON ―SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)‖ MARCH 26-27 2011
COS0102-4
Fig.5: Simulation results of excitatory and
inhibitory actions of postsynaptic Membrane. Top
waveform represents the EPSP and bottom
waveform represents the IPSP.
In an another model, they have used ion sensitive
field effect transistor (ISFET) as circuit analog and
incorporated into the famous Hodgkin-Huxley (H-H)
model of neuron to substitute the variable Na+ and
Cl- conductances, the details of which may be
obtained in reference [11]. They have mentioned
that such model has additional advantages over
MOSFET based model. The advantages that they
have high lighted are: (i) Measurement of different ions that diffuse through the post synaptic membrane and hence pH (ii) measurement of neurotransmitters diffused through the synaptic cleft by converting the ISFET into neurotransmitter sensitive enzyme modified FET (ENFET). This model, according to them, may become a useful research unit in neurology for biotelemetry applications. The second advantage that ENFET can also be used as circuit analog, in an effort, they have modeled it using ENFET sensitive to acetylcholine neurotransmitter for simulation of acetylcholine gated ion channels of the post synaptic membrane at the synaptic cleft[12].
MOSFET and ISFET based electrical models both
for excitatory and inhibitory actions of neurons
have been reviewed. It is concluded that ISFET
based models are more biologically motivated as
these are compatible with biological medium and
there is possibility of measurement of
neurotransmitters diffused through the synaptic
cleft by converting the ISFET into neurotransmitter
sensitive ENFET. These biologically motivated
models may become useful research and teaching
units in biomedical area in general and neurology in
particular.
REFERENCES [1] Hodgkin, A, L and Huxley, A. F., ―A
quantitative description of membrane current and its application to conduction and excitation in nerve‖, J. Physiol, 117. 500-544(1952)
[2] Hodgkin, A. L., ―Ionic movements and electrical activity in giant nerve fibers‖, Proceedings of the Royal Society of London. Series B, Biological Sciences, Vol. 148, 1-38(1957)
[3] Fitzhugh, R., ―Threshold and plateaus in the Hodgkin-Huxley nerve equations‖, J. Gen. Physiology, 43, 867-(1960)
[4] Johnson and Hanna, ―Membrane model: A single transistor analog of excitable membrane,‖ J. Theoret. Bio, 22, 401-411(1969)
[5] E.R. Lewis, ― Neuroelectric potentials derived from an extended version of the Hodgkin and Huxley model,‖ J. Theor. Biol. Vol.10,125-158, 1965
[6] L.D. Harmon and E.R.Lewis, ― Neural modelling”, Physiol. Rev. , Vol, 48, 513-591, 1966
[7] E.R.Lewis, ―Using electronic circuits to model simple neuroelectric interactions,‖ Proc.IEEE,vol 56, 931-949, June 1968.
[8 ] Xiao-lin Zhang, ―A Mathematical Model of a
Neuron with Synapses
based on Physiology‖, Nature Proceedings,
npre.2008.1703.1. March
2008.
[9] Soumik Roy, Jiten Ch Dutta, Shikhamoni
Phukan, ―Integrate-and-
Fire based Circuit model for simulation of
excitatory and inhibitory
synapses‖, Canadian Journal on Biomedical
Engineering & Technology
Vol. 1, No. 2 March 2010, 49-51. [10] Jiten Ch Dutta and Soumik Roy
―Biologically motivated Circuit model for simulation of excitatory and inhibitory synapses‖, Canadian Journal on Biomedical Engineering & Technology Vol. 1, No. 2 June 2010, 49-51.
[11] Jiten Ch Dutta and Soumik Roy, ―An Electronic Circuit Model for simulation of Synaptic Communication: The NEUROISFET for Wireless Biotelemetry‖, accepted for publication in the IEEE conf. on wireless communication, 24-25 Feb, 2011, BITS, MESRA
[12] Jiten Ch Dutta and Soumik Roy ―Biologically inspired Circuit model for simulation of Acetylcholine gated ion channels of the Postsynaptic membrane at synaptic cleft‖,
CONFERENCE ON ―SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)‖ MARCH 26-27 2011
COS0102-5
Proc. 2010 IEEE-EMBS Conference on Biomedical Engineering and Sciences, IECBES2010, Nov 30–Dec’ 02, 2010
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0103
Organ Failure Assessment in Malarial Patients
Using Artificial neural networks
Mr. Jay Kumar Pandey
Shri Ramswaroop Memorial College of Engineering and Management , Lucknow , U.P
Email id : [email protected]
Abstract
Health Care Management is one of the most important
and most important research areas of the new
millennium. The main purpose of this work was to analyze
the data on malaria patients in India using the artificial
neural networks such as Brain maker and statistical
analysis software (SAS). This data is known that SOFA
(Sequential Organ Failure Assessment) score and this
information useful in providing the condition of the organ
and based on this, the patient’s survival rate can be
estimated. In this analysis, the same individual SOFA
score of different organ systems (esp. the ones which are
used to calculate the overall SOFA score) for 753 patients
admitted in an hospital in India with malaria were trained
using artificial neural networks to provide better
predictions of the survival rate compared to the overall
SOFA scores. Using the statistical analysis tools like SAS,
the statistical aspects of this data was studied. Also, using
SAS, analysis was done on the data of the Indian malaria
patients and the interested outcomes were projected in the
figures at the end of this paper. The results show that the
artificial neural network turns out to be an efficient
predictor of survival rate and the results were
comparable to the SOFA scoring system where the same
variables used for calculating the overall SOFA score
were used for training the neural network
Keywords: Artificial neural networks, SOFA, malaria,
organ failure, statistical analysis
1. Introduction
Currently available prediction models such as the Acute
Physiology and Chronic Health Evaluation APACHE) II,
Simplified Acute Physiology Score (SAPS), and
Mortality Probability Models (MPM) use values taken
within the first 24 hrs of an ICU stay. However, these
scores ignore the many factors that can influence patient
outcome during the course of an ICU stay beyond the first
24 h. The Sequential Organ Failure Assessment (SOFA)
assesses patients for organ dysfunction not only at ICU
admission but serially APACHE II, SAPS, MPM etc,
researchers never stop working on during the ICU stay
and was first developed to evaluate morbidity.
Although this scoring system was developed to describe
and quantify organ function and not to predict outcome,
the obvious relationship between organ dysfunction and
mortality has been demonstrated in several studies. These
SOFA scores do not predict the exact rate of morbidity
every time but provide meaningful result most of the
times. Nowadays although we have many scoring systems
like creating a better system compared to the other which
always happens in a continually developing field such as
medicine. This work was done in an attempt to find such
system which is efficient compared to the standard system
like SOFA score.
Determining the patients survival chance based on these
scoring systems is very critical in an Indian hospital
environment. In this study two different methods of
predicting the hospital mortality were reviewed. One was
using these SOFA scores, doing the statistical analysis on
the available data to see how accurate are these scores in
predicting the mortality while the other one is using the
neural networks to predict the mortality rate. Later these
two results were compared to see which method is better
in predicting the mortality.
This work was done in an attempt to develop an
efficient method compared to the standard method. It was
observed that the neural networks were as efficient as the
traditional SOFA scoring system in predicting the
mortality of the patients.
1.1. Severe Malaria
Unless P. falciparum malaria is promptly diagnosed and
treated, the clinical picture may deteriorate rapidly. There
is significant morbidity and mortality associated with the
occurrence of severe malaria. Young children, pregnant
women, immunosupressed patients and any non-immune
persons are at risk for the development of complications.
One can assume that all South Africans living in the
malaria areas in this country and all South African
travelers are non-immune.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0103
1.2. Indicating Features
1.2.1. Clinical Features
Impaired consciousness, convulsions, Respiratory
distress: acidosis, ARDS, pulmonary edema, Jaundice,
Bleeding and Shock.
1.2.2. Biochemical Features
Renal impairment – serum creatinine, >265 μmol/liter or
rapidly rising creatinine or urine output
>265 <400 ml/day (adult)
• Acidosis (plasma bicarbonate <15 mmol/liter)
(serum lactate > 5 mmol/liter)
• Hepatic impairment
(transaminases > 3 times normal)
• Hypoglycemia (blood glucose < 2.2 mmol/liter)
• Hypoxia (pO2 - < 8 Kpa in room air)
1.2.3.Haematological Features
• Parasitaemia > 5%, or > 3+
• Haemoglobin < 6 g/l or haematocrit<20%
• > 5% neutrophils contain malaria pigment
• Presence of schizonts of P. falciparum in peripheral
blood smears
• Evidence of DIC
Figure.1. A.Cellular events that occur late in disease
development and contribute to malaria pathogenesis.
The early events of innate immune-cell and platelet
activation by parasite antigens, and the production of
cytokines, chemokines and vaso active molecules are not
depicted. P RBCs adhere through e CAMS
(i) Pselectinis expressed at low levels in ECs that
have not been activated, indicating that e CAMs with high
basal levels, such as ICAM1, and CD36 function in early
adhesion. This cell adhesion, together with molecules
derived from immune cells and platelets, activates ECs to
increase their e CAM levels, increasing p RBC adhesion
(i) and facilitating the adhesion of platelets
(ii) and leukocytes
(iii) All cell types use similar processes and cell-
adhesion molecules to adhere to the endothelium (i–iii).
They can also form rosettes in the blood, which are
analogous to mini-thrombi in sepsis. Cell adhesion and
vasoactive molecules released during infection cause
disruption of the endothelial tight junction
(iv) and, potentially, leakage of plasma proteins into
the interstitial space
(v). whether this leak becomes greater, or ECs
become apoptotic and slough off, leaving gaps for
endothelial hemorrhage
(vi), has yet to be determined. The blockade of
functional capillaries leads to a build up of lactic acid in
the tissue
(vii) but only mild hypoxia because the blood flow in
the arterioles is maintained during ECM.
Figure.1.B. Annex in V-stained endothelial cells
incubated with neutrophils plus serum from a patient with
fatal P. falciparum malaria (green fluorescence). Necrotic
cells were stained with propidium iodide (red
fluorescence). The inset shows endothelial cells incubated
with serum from a healthy person as a negative control.
Figure.1.B
Figure.1.C.
Figure.1.(C) TUNEL-stained kidney (a) and lung (b)
sections from two patients who had died from fatal P.
falciparum malaria showed the highest apoptosis rates
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0103
(median of 61% and range of 54 to 68 for two patients),
followed by sera from patients with severe P. falciparum,
mild P. falciparum, and P. vivax malaria. Sera from
Patients with fatal malaria plus neutrophils induced large
efects in the endothelial cell layers, a phenomenon that is
consistently seen secondary to apoptosis.
2. Observed Organ Failure Most of the patients with malaria have a pre-existing
organ dysfunction or failure or develop organ failure
during the progress of the disease sometimes which may
lead to death of the patient. Organ dysfunction is
associated with high rates of hospital morbidity and
mortality and as such accounts for a high portion of ICU
budget. Recently developed organ failure scores such as
Sequential organ failure Assessment (SOFA) can help
assess organ dysfunction or failure over time and are
useful to evaluate Morbidity.
Malaria occurs in 300–500 million individuals annually,
resulting in 1.5–2.7 million deaths. Most deaths occur due
to Plasmodium falciparum infection, which produces life
threatening cerebral, renal, hepatic, and hematological
dysfunction in about 1% of cases. This causes infected red
blood cells (RBCs) to adhere to capillary and venular
endothelium, non infected erythrocytes, and platelets,
resulting in circulatory obstruction. The brain is the most
common organ to be involved in severe malaria. Up to
45% of cerebral capillaries may be occluded at
postmortem examination because the receptors to which
infected RBCs adhere are maximally expressed on
cerebral capillary endothelium. Coma also may result
from interference with synaptic transmission by nitric
oxide from vascular endothelium, raised intracranial
pressure due to vasodilatation, and increased capillary
permeability. A combination of microcirculatory
occlusion, cytokine activation, and nitric oxide-mediated
changes in vascular tone are believed to cause organ
dysfunction that characterizes severe malaria. However,
malaria is not well recognized in critical care literature as
a cause of multiple organ dysfunction syndromes.
3. Neural Network based Model
Artificial neural networks are computational paradigms
based on mathematical models that unlike traditional
computing have a structure and operation that resembles
that of the mammal brain. Neural networks lack
centralized control in the classical sense, since all the
interconnected processing elements change or “adapt”
simultaneously with the flow of information and adaptive
rules. The commonest learning mechanism in artificial
neural networks is the back-propagation algorithm,
wherein the system predicts the outcome for each patient
based on past experience (memory) and compares this
with actual outcome.
The advantage of neural networks over conventional
programming lies in their ability to solve problems that do
not have an algorithmic solution or the available solution
is too complex to be found. Neural networks have been
applied within the medical domain for clinical diagnosis,
image analysis and interpretation, signal analysis and
interpretation, and drug development.
4. Data Analysis and Knowledge Discovery
The development phase of the neural network began with
the conception of a study protocol which involved the
comparison of the morbidity rates predicted by APACHE
II and Artificial neural networks. Several models of
artificial intelligence techniques have been used in the
ICU. One such technique suited to predict mortality is the
artificial neural network.
The main goal of this paper is to compare neural
networks with an already validated and commonly used
outcome prediction model. The question remains whether
a model such as SOFA scoring is better at predicting
hospital outcome than a model derived from Indian
patients treated in an Indian hospital. It was therefore
attempted to compare the predictive accuracy of artificial
neural networks derived from Indian patients with the
SOFA scoring system.
The variables which were used to train the neural
network were similar to the ones used for predicting the
overall SOFA score. The sofa scores used were those of
the six different organ systems estimated during the first
48hrs of the patient’s admission in the hospital.
The network training was done using Brain maker
professional and Net maker Professional. The data
which has been worked on was that of 753 patients who
were suffering from malaria in an Indian hospital. Out of
this, data on 553 patients was used to train the network
and that of 200 patients was used to test the network. The
results were then read in to a text file.
Using SAS (Statistical Analysis Software), the
frequency bar charts was drawn for survival and SOFA
scores. The SOFA Scores were then recoded to six
different ranges and the table analysis was done with
these ranges and survival. This table analysis gives the
percentage of patient’s corresponding to each different
SOFA range. The correlation procedure was done to find
the correlation between SOFA score and survival.
Using Excel 2002, graphs were drawn for the output
acquired from neural network to see the percentage of
people survived and also graphs were drawn on these 200
patients used for testing the neural network to see the
actual percentage of people survived. Then both of these
results were compared to see predictive capability of the
trained network.
Using Access 2002, queries were written to compare the
real survival rate with that of the output acquired from
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0103
neural networks. Also data was queried to find if the OFA
score ranges actually predict the mortality rate
comparable to that of the real data. Then those data of
patients whose SOFA range is not consistent with the
survival/mortality rate were compared to the prediction
done by ANN’s and it was found that the survival rate for
these patients was predicted accurately by ANN’s.
Figure .2.A. Sequential Organ Failure Assessment Data
5. Experimental Results Using Neural Networks:
After training the data using Brain maker, the least
RMS error and Average errors were determined and the
plot of the output is shown in the figure 2.
Some of the data were used to validate the model. The
trained model behaved very well and the accuracy of the
model was over 88%. It was found that the performance
of the model of neural networks was significantly better
than that of the SOFA score model when applied to the
given data set.
The SOFA range 4 had the patients who both survived
as well as died which explains that SOFA scoring system
also has its inadequacies. When this SOFA range (SOFA
range=4) was further analyzed using Microsoft Access
2002 to check the individual values present in this range,
it was observed that 3 patients survived and 3 patients
died. Surprisingly, it was also observed that the 3 patients
out of which 2 people survived and 1 died had the same
SOFA score of 16. This doesn’t explain the fact how 2
persons with the same SOFA score can show distinct
Survival capabilities. When these 6 patients who had the
SOFA range of 4 was compared to the results predicted
by Brain maker, it was observed that the neural network
(Brain maker) gave the correct prediction rates for these
patients.
The correlation procedure was performed on the test
data (200 patients) on SOFAMAX and the SURVIVAL.
It was observed that the parameters have very good
correlation as seen in figure3.
These SOFAMAX values were then recoded in to
ranges to better predict the variations in survival for
different SOFA score ranges. The table analysis of these
SOFAMAX ranges with SURVIVAL was done indicating
the percent of patients in each SOFAMAX range as
shown in figure 4.
Figure 2 .B. Statistical analyses of the data using SAS:
Figure 3: Correlation Procedure
Figure 4: Data analysis of SURVIVAL and
SOFAMAX ranges The histograms for these SOFAMAX ranges was
plotted for both survivors and non-survivors indicating
the total number of survivors/non survivors for each
SOFAMAX range as shown in figure 5. The frequency
plot of the non-survivors with different SOFA score
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0103
ranges is supposed to show that the percentages of people
with higher SOFA score are supposed to have higher rate
of mortality.
Figure 5: Bar Chart of survivors and non-survivors
for different SOFAMAX ranges
But it’s observed that there is a decline in the graph
level which suggests that the rate of mortality is lesser in
the SOFA range-6 compared to the Percentage of
mortality with SOFA range 5. This percentage decline
with increase in SOFA range leads to an ambiguous
condition. When results of the tested network were
checked with the original data, it was observed that the
trained network produced accurate results for 198 patients
out of these 200 patients, which shows that it gave 99% of
the exact results. The reason for being so accurate in
prediction might have the possibility that the data used for
training and testing is very low. Most of the times in the
real world situations the data sets may be so large that the
neural network is not trained as good as it was in this case
and also the results may not be as accurate as it was in
this case.
6. Conclusion
This project was aimed at making the peoples life in
Medical field (like physicians, nurses) easier while
assisting by providing them with tools which have been
created with the improving technology to make the
complex decisions easier. Such systems not only help
them to predict the outcomes but are much efficient in
cutting costs of the hospital due to early discovery of this
knowledge. This project also aims at such embedded
technology into the field of medicine to make the
complex decision easier and foresee the outcome most of
the times.
This work has been successful in achieving its
objective. The results are reliable. Based on the results
above still it can be said that it is an efficient neural
network which works as good as the standard SOFA
scores and is also little efficient compared to the standard
SOFA scoring system here.
7. References [1] http://www.malaria.org/
[2] http://www.ncbi.nlm.nih.gov/entrez/query.fcgi
[3] http://www.cdc.gov/malaria/
[4] http://my.webmd.com
[5] http://home.uchicago.edu/~junji/KRL_HP/chestrad.htm
[6] www.nd.com/whatisnn.htm
[7] http://www.openclinical.org/home.html
[8] “Intelligence in medical image processing” Image and
Vision Computing, 19(4), 177.
[9] Dinesh Mital, Shankar Srinivasan, Syed Haque and Radhika
Uppalapati Dept. of Health Informatics Univ. of Medicine and
Dentistry of New Jersey School of Health Related Professions
Newark, NJ.
[10] Aizenberg I., Aizenberga N, Hiltnerb J et al (2001).
“Cellular neural networks and computational intelligence
inmedical image processing.” Image and Vision Computing,
19(4), 177-183.
[11] Anthony, D., Hines, E., Barham, J., & Taylor, D. (1990).
“A Comparison of Image Compression by Neural Networks and
Principal Component Analysis”. In International Joint
Conference on Neural Networks, Vol. 1, pp. 339--344.
[12] Baxt, W. G. (1995). “Application of Artificial Neural
Networks to Clinical Medicine”. Lancet, 346, 1135-1138.
[13] “Review of Neural Network Applications in Medical
Imaging and Signal Processing.” Medical and Biological
Engineering and Computing, 30(5), 449-464.
[14] Miller, A. (1993). “The Application of Neural Networks to
Imaging and Signal Processing in Astronomy and Medicine”
Ph.D. thesis, Faculty of Science, Department of Physics,
University of Southampton.
[15] Weinstein, J., Kohn, K., Grever, M., et al. (1992). “Neural
Computing in Cancer Drug Development: Predicting
Mechanism of Action”. Science, 258, 447--451.
[16] Weinstein, J., Kohn, K., Grever, M., et al. (1992). “Neural
Computing in Cancer Drug Development: Predicting
Mechanism of Action”. Science, 258, 447--451.
pp. 3546--3551.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-1
AN INTELLIGENT CONTROL METHOD TO REDUCE BRAKE NOISE
Pramod Kumar Pandey
Department of Electrical Instrumentation
Engineering Thapar University Patiala
Punjab -147004
Yaduvir Singh
Department of Electrical Instrumentation
Engineering Thapar University Patiala
Punjab -147004
Abstract: In this paper, a novel approach
to suppress vibration that causes brake
noise is proposed employing a closed-loop
feedback control method using an active
force control (AFC) based strategy and
fuzzy logic controller. The idea is to
introduce an active element that
dynamically compensates the
disturbances through a control
mechanism that takes into account the
direct measurements and estimation of
parameters in the AFC section. A disc
brake model is considered and simulated
taking into account a number of
operating and loading conditions. Results
clearly show the superiority of the
proposed AFC + fuzzy -based scheme
compared to the AFC+ PID and pure PID
counterpart in suppressing the vibration
and hence the brake noise.
Key Terms - Brake noise; vibration;
robust; active force control, PID, Fuzzy
logic controller.
I. INTRODUCTION
The brake system of an automobile
typically consists of the contact metallic
solids rubbing against each other, which
frequently generates undesirable noise and
vibrations. Thus, noise generation and
suppression have become an important
factor to be considered in the design and
manufacture of brake components. Indeed,
as note by Abendroth and Wernitz [1], a
large number of manufacturers of brake pad
materials spend up to 50% of their
engineering costs on issues related to noise,
vibration and harshness. Even to this day,
there is no precise or conclusive definition
of brake squeal that has gained complete
acceptance. It is also worth mentioning that
since in a vehicle with disc brakes installed
at the front wheels while drum brakes at the
back wheels, around 70% of the braking
action occurs at the front wheels. Thus, it is
expected that most of the noise and squeal is
coming from the front disc brake system.
In literature, there are three major
methods to study and reduce brake squeal,
namely through mathematical modeling,
experimental and finite element methods.
One of the most recent works that has been
carried out for reducing brake noise using
finite element (FE) can be found in [3].
They developed a dynamic FE model of the
brake system, and based on their analysis,
the pad design changes can be made in the
FE model to determine the potential
improvements in the dynamic stability of the
system and also in noise reduction. Wagner
et al. proposed a new mathematical rotor
based model of a brake system that is
suitable for noise analysis [4]. Brief
descriptions of the previous mathematical
models that have been developed by other
researchers were explicitly outlined in their
study. Besides, there is also an active control
method known as dither control which
makes use of high frequency disturbance
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-2
signal for the suppression of the automotive
disc brake squeal. Through this scheme, the
dither signal stabilizes friction induced self-
oscillations in the disc brake using a
harmonic vibration, with a frequency higher
than the squeal frequency generated from a
stack of piezoelectric elements placed in the
caliper piston of the brake system. The
resulting control vibration was not heard
from the brake system if an ultrasonic
control signal was activated. This system
assumes an open loop control mode in
which there is no requirement to detect the
presence of squeal and is much simpler in
design than the feedback control [5]. This
paper presents a closed loop control
employing system Active Force Control
(AFC) with PID element applied to a brake
model described in [6] in order to suppress
the brake noise and squeal. The main
advantage of the AFC technique is its ability
to reject disturbances that are applied to the
system through appropriate manipulation of
the selected parameters.
In addition, the technique requires much
less computational burden and has been
successfully demonstrated to be readily
implemented in real-time. AFC as first
proposed by Hewitt and Burdess [7] is very
robust and effective in controlling a robot
arm. Mailah [8, 9] has successfully
demonstrated the application of the
technique to include many other dynamical
systems with the incorporation of artificial
intelligence (AI) methods.
In this paper fuzzy logic+AFC based
model shows the better result than PID and
PID+AFC model. It also provides better
control accuracy and fast speed of response.
II. THE BRAKE MODEL
A disc brake system assuming a two
degree of freedom model based on the one
described in [6] is considered in the study.
The model consists of a conveyor belt with
constant velocity vB that is pushed with a
constant normal force FN against a block
modeled as a block of mass m. Fig. 1 shows
that the model is just a single-point mass
sliding over a conveyor belt and there are
two linear springs k1 and k2 parallel and
normal to the belt surface with the latter
regarded as the physical contact stiffness
between the objects in relative sliding
motion. In addition, there is another linear
spring k mounted at oblique angle of 45°
constituting the off-diagonal terms in the
model‟s stiffness matrix. For the friction
component, a coulomb model is assumed
such that FT = μ FN, where μ is the
coefficient of kinetic friction usually taken
to be constant. FN is a normal force and
since the normal force at the friction
interface is linearly related to the vertical
displacement x2 of the mass then the
resulting friction will become FF = μ k2 x2.
Assuming that the mass of the conveyor belt
system is larger than the mass block, it
implies that the vibration of the belt does not
show any changes due to its inertia.
The matrix form of the equation of motion
can be expressed as [6]
1 1 1 2 2
2 2 2
1 1 1
2 2 2
0 / 2 / 20
0 / 2 / 20
x x x k x
x x x N
c k k km
c k k km
----------------- (1)
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-3
Figure 1.Two DOF model of a disc brake
system
III. CONTROL STRATEGY
After acquiring the model of the disc
brake and its related equation of motion, it is
required to control the vibration of the mass
with respect to the vertical direction, x2
considering an actuator that produces a force
F in parallel with the preload force N. Thus,
(1) will be written as
1 1 1 2 2
2 2 2
1 1 1
2 2 2
0 / 2 / 20
0 / 2 / 20
x x x k x
x x x N F
c k k km
c k k km
-------------------- (2) A robust control strategy is proposed
here employing an Active Force Control
(AFC) based scheme that is used in
conjunction with the conventional PID
controller. The PID controller was first
tuned with Ziegler-Nichol‟s method for
good performance and later the AFC part
was incorporated into the system to provide
the compensation of the disturbances. Fig. 2
shows the AFC scheme applied to a
dynamic translation system (disc brake).
AFC scheme is shown to be very effective
provided the actuated force and body
acceleration are accurately measured and at
the same time the estimated mass property
approximated [7, 8]. The essential AFC
equation can be related to the computation
of the estimated disturbance Fd as follows:
Fd= F- M’. a ------------- (3)
Where F is the measured actuating force,
M’ is the estimated mass and a is the
measured linear acceleration. This parameter
is then fed back through a suitable inverse
transfer function of the actuator to be
summed up with the PID control signal. The
theoretical analysis including stability of the
proposed AFC method has been sufficiently
described in [10]. A number of methods to
estimate the mass have been proposed in
previous studies such as through the use of
artificial intelligence (AI) and crude
approximation techniques [7-9]. In this
study, the use of crude approximation
method to approximate the estimated mass is
deemed sufficient. The main challenge of
the AFC method is to acquire appropriate
estimation of the mass needed to compute
the disturbance Fd in the feedback loop. A
conventional PID that is used with the AFC
scheme can be typically represented by the
following equation:
Gc(s) = Kp + Ki /s + Kd s -------------- (4)
Where Kp, Ki and Kd are the proportional,
integral and derivative gains respectively.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-4
Figure 2 Schematic diagram of AFC
strategy
IV. SIMULATION
The actuator is assumed to be a linear
type with a suitable constant gain. It
provides the necessary external energy to
suppress vibration in the model. The
parameters used in this study were taken
from the previous research [11]. However,
some of them need to be modified to suit the
application in the simulation. The detailed
parameters are as follows:
Minimal disc brake model parameters:
Body mass, m = 0.7 kg,
Spring stiffness, k = 10 N/m, k1 =
11 N/m, k2 = 20 N/m
Damping coefficient, c1 = 0.4 Ns/m,
c2 = 0.4 Ns/m
Friction coefficient, μ = 0.3
Normal preload, N = 5 N
Actuator:
Actuator gain, Q = 0.5
Reference value:
Reference input = 0.00 m (i.e. no
vibration)
Disturbance:
Magnitude of step function = 1 N
Frequency = 1.5 Hz
In this work, namely the step
disturbances are deliberately introduced to
the disc brake system to evaluate the
robustness of the system. The Simulink
diagram of the passive disc brake system
model is shown in Fig. 3. The schematic
block diagram was constructed from (1).
Figure 3 Passive brake system
In order to have an active disc brake
system, an actuator force for compensating
the disturbance force is required, and the
actuator force is controlled by a PID
controller which typically involves a
negative feedback loop. Hence, there are
two inputs to the dynamic disc brake system
which is the step disturbance and the
actuator force input. Figure 4 shows the
active disc brake system.
Figure 4 PID active brake control scheme
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-5
To achieve better overall performance
of the brake system, AFC is „added‟ to the
PID controller, The AFC Simulink blocks
includes the estimated mass, parameter 1/Q
and the percentage of AFC gain. The input
to the AFC control is the mass acceleration
and the output is summed with the PID
controller output and then multiplied with
the actuator gain which finally generated the
actuator force. In order to get the effective
results using this method, it is required to
have a suitable mass estimation combined
with the best tuning of the PID controller
gains. Also a memory block is used to
eliminate the algebraic loops and algebraic
variable problems. Figure 5 shows an AFC
scheme that has a step input as its
disturbance.
Figure 5 PID+ AFC model for brake control
To tune the PID controller, we use the
Ziegler-Nichol‟s method and the results are
tabulated as shown in Table-1 TABLE 1 PID PARAMETERS TUNED
USING ZIEGLER-NICHOL‟S METHOD
PID Kp Ki Kd Ti Td
Gain 0.6 0.857 0.105 0.7 0.17
The estimated mass for the AFC loop was
obtained by trial and error (crude
approximation method) in which a suitable
value was easily found to be 1.4 kg, and the
percentage of AFC used is 100% implying
that the AFC loop employs full AFC
implementation.
Model for AFC+Fuzzy
FLC provides an algorithm which can
convert the linguistic control strategy based
on expert knowledge into an automatic
control strategy. Experience shows that the
FLC yields results superior to those obtained
by conventional control algorithms. In
particular, the methodology of the FLC
appears very useful when the processes are
too complex for analysis by conventional
quantitative techniques or when the
available sources of information are
interpreted qualitatively, inexactly, or
uncertainly. Thus fuzzy logic control may be
viewed as a step toward a rapprochement
between conventional precise mathematical
control and human-like decision making.
In this research paper a fuzzy logic
controller is developed. The inputs to the
FLC are applied force and change in force.
The output is change in vibration.
Figure 6: Membership function for input-1
Figure 6 shows the membership
function for force.
Figure 7: Membership function for input-2
Figure 7 shows the membership
function for change in force.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-6
Figure 8: Membership function for output
Figure 8 shows the membership
function for change in vibration.
Figure 9: The rule viewer of fuzzy logic
controllers
Figure 9 shows the rules of fuzzy
logic controller.
Figure 10: 3-D surface view of fuzzy logic
controller
Figure 10 shows the three dimensional
surface view of input and output
membership functions of fuzzy logic
controller. Figure 11 shows the fuzzy based
control of vibration.
Figure 11: Fuzzy+AFC model for break
control
V.RESULTS AND DISCUSSION
After tuning the PID control system
and obtaining suitable values for other
relevant parameters, the simulation was
executed for a period of 5 s since it is usual
that the brake process does not take more
than that duration. At first, the step
disturbance is applied and then the
simulation was performed without using
AFC and only the pure PID controller was
applied. The result of this process can be
seen in Figure 12. It can be observed that the
vibration that may result in producing noise
is relatively high (a peak of more than 0.6 m
in amplitude in some regions). Next the
simulation was carried out again but this
time considering 100% AFC mode plus the
PID controller.
Figure 12: Response of the brake model
with a step input disturbance for system with
PID
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-7
Figure 13: Response in frequency domain
for PID
The result of the scheme is
superimposed onto previous graph obtained
as shown in Figure 11. It is evident that the
vibration is virtually reduced to a very low
value, implying that squealing can be almost
if not totally avoided. A closed-up view of
the graphical result of the PID+AFC scheme
is depicted in Figure 14. It clearly shows the
superiority of the scheme in rejecting the
vibration of the system. The maximum
amplitude of the vibration was less than
0.0004 m, which is very much lower
compared to the brake system operated with
a PID only controller.
Figure 14 Performances of PID+AFC
Controller
Figure 15: Amplitude vs. time plot using
step disturbance in AFC+Fuzzy
Figure 15 clearly shows the
superiority of the Fuzzy+AFC scheme in
rejecting the vibration of the system. The
maximum amplitude of the vibration was
much lower compared PID+AFC model.
Figure 16: Response of the brake model
with a step input disturbance for system with
AFC+Fuzzy
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0201-8
Figure 17: Response in frequency domain
for AFC+Fuzzy
VI. CONCLUSION
A novel AFC-based scheme has been
proposed to suppress the vibration and noise
(squeal) emanating from a disc brake
system. Based on the simulation results, it is
obvious that when a pure PID controller is
applied to the brake system, vibration and
noise are in fact reduced but still a
noticeable amount of them remain.
However, upon applying the AFC+PID the
vibration and noise (squeal) are significantly
reduced and approaching zero datum.
Fuzzy+AFC strategy is robust and effective
in countering the undesirable effects. Future
works may include using neuro-fuzzy
algorithm with AFC model.
REFERENCES [1] H. Abendroth and B. Wernitz, “The
integrated test concept: dynovehicle,
performance-noise”, Technical Report 2000-01-
2774, SAE, Warrndale, PA, 2000.
[2] N.M. Kinkaid, O.M. O‟Reilly, and P.
Papadopoulos, “Review: Automotive disc brake
squeal”, Journal of Sound and Vibration, vol.
267, pp. 105-166, 2003.
[3] Yi Dai, Teik C. Lim, “Suppression of brake
squeal noise applying finite element brake and
pad model enhanced by spectral-based assurance
criteria”, Applied Acoust (2007), doi: 10.1016 /
j.apacoust. 2006.09.010
[4] Utz von Wagner, Daniel Hochlenert and
Peter Hagedorn “Minimal Models for Disk
Brake Squeal”, Journal of Sound and Vibration,
vol. 302, pp. 527-539, 2007.
[5] A. Grag, “Active control of automotive disc
brake rotor squeal using dithers”, Master Thesis,
Georgia Institute of Technology, 2000.
[6] N. Hoffmann, N. Wagner, and L. Gaul,
“Quenching mode-coupling friction-induced
instability using high-frequency dither”, Journal
of Sound and Vibration, vol. 279, pp. 471–480,
2005.
[7] J.R. Hewit and J.S. Burdess, “Fast dynamic
decoupled control for robotics using active force
control”, Trans. on Mechanism and Machine
Theory, vol. 16, no.5, pp. 535-542, 1981.
[8] M. Mailah, “Intelligent active force control
of a rigid robot arm using neural network and
iterative learning algorithms”, Ph.D Thesis,
University of Dundee, UK, 1998.
[9] M. Mailah and N.I.A. Rahim, “Intelligent
active force control of a robot arm using fuzzy
logic”, in: Proceedings of IEEE International
Conference on Intelligent Systems and
Technologies, Kuala Lumpur, Malaysia, vol. 2,
pp. 291-296, 2000.
[10] J.S. Burdess, J.R. Hewit “An active method
for the control of mechanical systems in the
presence of unmeasurable forcing” Mechanism
and Machine Theory, vol. 21, no. 5, pp. 393-
400, 1986.
[11] N. Hoffmann, M. Fischer, R. Allgaier, and
L., Gaul, “A minimal model for studying
properties of the mode-coupling type instability
in friction induced oscillations”, Mechanics
Research Communications, vol. 29, pp. 197-205,
2002.
[12] S.M Hashemi-Dehkordi, M. Mailah and
A.R. Abu-Bakar “A Robust Active Control
Method to Reduce Brake Noise” Proceedings of
the 2008 IEEE International Conference on
Robotics and Biomimetics Bangkok, Thailand,
2009 978-1-4244-2679-9/08/ ©2008 IEEE.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0202-1
APPLICATION OF FUZY LOGIC IN AUTOMATIC BOTTLE FILLING SYSTEM
Nikita Agarwal1, Ritika Srivastava2, Preeti Dhiman3
EIE Department, Galgotias college of Engineering And Technology, Greater Noida
Abstract : In this paper we described the realization of an automatic Bottle Filling system designed with Fuzzy logic and implemented with microcontroller and matlab program .Previously the system was semiautomatic ,the major downfall of this approach was that system was not able to impose correct pressure inside the bottle ,so there were often Jamming condition. The fuzzy Logic method overcomes these shortcomings by deploying a set of heuristic rules and procedures derived from the deep knowledge obtained from experienced skilful control engineers. In this method we uses the Proximity sensor and Light dependent resistor to detect the presence of bottle and level of liquid inside it and Fuzzy control is applied to regulate the level of liquid inside the bottle.
Keywords: Fuzzy Logic, Heuristic rules, Fuzzy control, automatic bottle filling production, matlab, proximity sensor.
I. INTRODUCTION
Automatic Bottle Filling systems are used on large scale in Industries such as mineral water, soft drinks, food beverages and other similar industries. According to the International bottled water association (IBWA), sale of Bottled water has been tremendously increased by 500 percent over the last decade [1].The bottling procedure has been automated from a long time. Here we developed a proximity sensor using a infrared sensor that detects the presence of bottle and light dependent resistor and comparator is employed to sense the level of liquid in bottle.
Nikita Agarwal1, B.tech (Final year) , is with the Electronics and Instrumentation department, Galgotias College Of Engineering and Technology ,Greater Noida ,India (corresponding author to provide phone: 9953197920; email: [email protected])
Ritika Srivastava2 , Btech( Final year) , is with the Electronics and Instrumentation Department ,Galgotias college of Engineering and Technology ,Greater Noida ,India(corresponding author to provide phone: 9452829482 ; email: [email protected]).
Preeti Dhiman3 is working as Assistant Professor in EIE Department, GCET, Greater Noida, India(email:[email protected], phone: 9911923434)
Fuzzy logic methodology is applied to regulate the level of liquid. Microcontroller provides the sample input to the computer and with these input Matlab program generates control signal for solenoid tab. We apply Fuzzy principles to a real plant , to see if an automated fuzzy could improve the performance of plant, avoiding bottle to get stuck .The outline of bottle Filling system is shown in figure 1.
Fig 1. Outline of bottle filling system Courtesy: OMRON Tech for Focused Automation [2]
Fig2.Over and under filled Bottle Courtesy: OMRON Tech for Focused automation [2]
II. LITERATURE REVIEW
The semiautomatic system is not able to impose the right pressure level inside the different section of frame to maintain a constant velocity of the bottles along the production line, which often resulted in jamming condition. In [3] E.Mainardi used fuzzy logic for realization of air transportation bottling plant controller. The use of Fuzzy logic for
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0202-2
the design and control of production system is widely used and recognized by scientists. Some studies described recommendation for effective application of Fuzzy Logic in production system. Fuzzy approach is also used in part family formation [4] , control of an industrial desktop robot which are used in different industrial application such as dispensing , soldering , screw tightening[5]. Recently fuzzy logic controller is employed in developing modular and reconfigurable robots (MRR) in flexible automation to reduce labour and increase throughput [6].Our study will focus on analyzing a fully automatic bottle filling system with respect to performance measures, equipment failures and breakdown.
III BRIEF DESCRIPTION OF THE SYSTEM
The system under study is a fully automatic bottle filling system. It consists of a sensor to detect the level of the liquid, RS232 and MAX232 is used for interfacing of microcontroller with computers, microcontroller to detect and transfer the signal and finally it is implemented using fuzzy logic and matlab programs. The system layout is shown in figure 3.
Fig 3. System layout.
A.AT89S52
The 89S52 is a low-power, high-performance CMOS 8-bit microcontroller with 8K bytes of in-system programmable Flash memory. The device is manufactured using Atmel’s high-density non-volatile memory technology and is compatible with the industry-standard 80C51 instruction set and pinout. The on-chip Flash allows the program memory to be reprogrammed in-system or by a conventional non-volatile memory programmer. By combining a versatile 8-bit CPU with in-system programmable Flash on a monolithic chip, the embedded control applications. The AT89S52 provides the following standard features: 8K bytes of Flash, 256 bytes of RAM, 32 I/O lines, Watchdog
timer, two data pointers, three 16-bit timer/counters, a six-vector two-level interrupt architecture, a full duplex serial port, on-chip oscillator, and clock circuitry. In addition, the AT89S52 is designed with static logic for operation down to zero frequency and supports two software selectable power saving modes. The Idle Mode stops the CPU while allowing the RAM, timer/counters, serial port, and interrupt system to continue functioning. The Power-down mode saves the RAM con-tents but freezes the oscillator, disabling all other chip functions until the next interrupt. The connections are shown in fig 4.
Fig 4. AT89S52 connection
B. MAX232
The MAX232 is an integrated circuit that converts signals from an RS-232 serial port to signals suitable for use in TTL compatible digital logic circuits. The MAX232 is a dual driver/receiver and typically converts the RX, TX, CTS and RTS signals. The drivers provide RS-232 voltage level outputs (approx. ± 7.5 V) from a single + 5 V supply via on-chip charge pumps and external capacitors. This makes it useful for implementing RS-232 in devices that otherwise do not need any voltages outside the 0 V to + 5 V range, as power supply design does not need to be made more complicated just for driving the RS-232 in this case. The receivers reduce RS-232 inputs (which may be as high as ± 25 V), to standard 5 V TTL levels. These receivers have a typical threshold of 1.3 V, and a typical hysteresis of 0.5 V. The later MAX232A is backwards compatible with the original MAX232 but may operate at higher baud rates and can use smaller external capacitors – 0.1 μF in place of the 1.0 μF capacitors used with the original device. The corresponding figure is shown in fig 5.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0202-3
Fig 5. MAX232 connection
A. RS232
In RS-232, user data is sent as a time-series of bits. Both synchronous and asynchronous transmissions are supported by the standard. In addition to the data circuits, the standard defines a number of control circuits used to manage the connection between the DTE and DCE. Each data or control circuit only operates in one direction that is, signaling from a DTE to the attached DCE or the reverse. Since transmit data and receive data are separate circuits, the interface can operate in a full duplex manner, supporting concurrent data flow in both directions. The standard does not define character framing within the data stream, or character encoding. The corresponding figure is shown in figure 6.
Fig 6. RS232 connection
C. ULN2003
The ULN2003 is a monolithic high voltage and high current Darlington transistor arrays. It consists of seven NPN Darlington pairs that feature high-voltage outputs with common-cathode clamp diode for switching inductive loads. The collector-current rating of a single Darlington pair is 500mA. The Darlington pairs may be paralleled for higher current capability. Applications include relay drivers, hammer drivers, lamp drivers,
display drivers (LED gas discharge), line drivers, and logic buffers. The ULN2003 has a 2.7k series base resistor for each Darlington pair for operation directly with TTL or 5V CMOS devices. The diagram of ULN 2003 is shown in figure 7.
Fig 7 . ULN2003 connection
B. PROXIMITY SENSOR
A proximity sensor is a sensor able to detect the presence of nearby objects without any physical contact. A proximity sensor often emits an electromagnetic or electrostatic field, or a beam of electromagnetic radiation (infrared, for instance), and looks for changes in the field or return signal. The object being sensed is often referred to as the proximity sensor's target. Different proximity sensor targets demand different sensors. For example, a capacitive or photoelectric sensor might be suitable for a plastic target; an inductive proximity sensor requires a metal target. The maximum distance that this sensor can detect is defined "nominal range". Some sensors have adjustments of the nominal range or means to report a graduated detection distance. Proximity sensors can have a high reliability and long functional life because of the absence of mechanical parts and lack of physical contact between sensor and the sensed object. Proximity sensors are also used in machine vibration monitoring to measure the variation in distance between a shaft and its support bearing. This is common in large steam turbines, compressors, and motors that use sleeve-type bearings.IEC 60947-5-2 define the technical details of proximity sensors. A proximity sensor adjusted to a very short range is often used as a touch switch. A
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0202-4
proximity sensor is divided in two halves and if the two halves move away from each other, then a signal is activated. A proximity sensor can be used in windows, and when the window opens an alarm is activated. The figure is shown in fig 8.
fig 8. The design of infrared proximity sensor courtesy: Autonics sensor and controller [7]
IV DEVELOPMENT AND IMPLEMENTATION OF FUZZY LOGIC IN THE SYSTEM
FUZZY LOGIC:
Fuzzy logic is an approach to computing based on "degrees of truth" rather than the usual "true or false" (1 or 0) Boolean logic on which the modern computer is based. The idea of fuzzy logic was first advanced by Dr. Lotfi Zadeh of the University of California at Berkeley in the 1960s. Dr. Zadeh was working on the problem of computer understanding of natural language. Number of partial truths which we aggregate further into higher truths which in turn, when certain thresholds are exceeded, cause certain further results such as motor reaction. A similar kind of process is used in artificial computer neural network and expert systems. It may help to see fuzzy logic as the way reasoning really works and binary or Boolean logic is simply a special case of it.
The implementation of fuzzy logic is shown in figure 9,10 11 ,12 and 13. The input is taken as Level , abbreviated as L1 ,L2 and L3. The Corresponding output is taken as Valve Opening.
Fig 9. Bottle filling system with level as input and valve opening as output
Fig 10. Membership function plot of input variable “Level”
Fig 11. Membership Function plot of Output variable “valve opening”
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0202-5
Fig 12. Rules of bottle filling system
Fig 13. Rule viewer of bottle filling system
V. POSSIBLE PROBLEMS ON BOTTLING PLANT
During transportation of Bottles several problem can occur , depending on shape and size of bottle , environmental conditions (temperature and humidity can affect the sliding of bottle).For instance one bottle gets stuck on the guide then it blocks all other bottles which are behind it. The only solution in this situation is manual intervention of plant supervisor. But during the standard working of plant often the supervisor has to change the pressure level of the pipe to maintain a constant bottle flow. But sometimes operators do not know what to do in particular situation. This is due to the complexity of plant .It is clear that this is the typical fuzzy situation, so we tried to design a fuzzy controller to modify the
pressure levels in order to maintain a constant speed and flow of Bottle in entire plant and avoiding Jamming Condition.
VI. CONCLUSION
The study shows how Fuzzy Logic is used for designing the automatic bottle filling system. Bottled water is the fastest growing beverage industry in the world. A proximity sensor is employed for detecting the presence of bottle and Light dependent resistor and comparator is employed for sensing the level of liquid inside the bottle. Fuzzy Logic is employed for regulating the liquid level. There is no new investment on machine , conveyors, or anything else.
REFERENCES
[1] WATER PET BOTTLE, From trash to treasure, august 2001 .available at http://www.designboom.com
[2] OMRON Tech Focused for automation solution.
[3] E.Mainardi, M.Bonfe, M.Golfieri, “Fuzzy controller for an air conveyor bottling plant”, 16
th
IEEE conference on control applications, Singapore 1-3 0ctober, 2007
[4]Chwen –Tzeng Su,” A Fuzzy Approach For Part Family Formation “, International IEEE, page 289 (1995).
[5] Santos, CM.P; Ferreira, M.J,” Control Of an Industrial desktop Robot Using computer Vision and fuzzy rules”, IEEE International , Volume 3 ,page 1297-1303, 2005.
[6] Biglarbegian, M; Melek, W; Mendel, M;” Design of Novel Interval type- 2 Fuzzy Controller For Modular and reconfigurable robots: Theory and Experiments” IEEE, Volume pp,Issue:99, page (1),2010.
[7] Autonics sensor and controller. Available at http://www.autonics.com.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-1
Real Time Control of Rotary Flexible Joint with LQR and Fuzzy
Controller
Mr. Jay Kumar Pandey
Shri Ramswaroop Memorial College of Engineering and Management , Lucknow , U.P
Email id : [email protected]
Abstract
In the position control of robot arm, Rotary
Flexible Joint (ROTFLEX) constitutes main
function of positioning the arm. The precision
with which the arm is controlled depends on the
response of motor and on different parameters
of flexible joint. This paper concentrates on
development of Linear Quadratic Regulator
(LQR) and fuzzy controller for flexible joint
operated by a dc servomotor. The comparison of
performance of real time system with simulation
is done by simultaneously observing measured
and simulated angular position of flexible joint.
It is found that the response is improved with
fuzzy controller as compared to LQR.
1. Rotary Flexible Joint (Rotflex)
The rotary flexible joint consists of a rotary
sensor mounted in a solid aluminum frame and
is designed to mount to a Quanser rotary servo
plant. The sensor shaft is aligned with the motor
shaft. One end of a rigid link is mounted to the
sensor shaft. The link rotation is counteracted by
two extension springs anchored to the solid
frame resulting in an instrumented flexible joint.
The spring anchor points are adjustable to three
locations to obtain various stiffness constants.
Three types of springs are supplied with the
system resulting in a total of 9 possible stiffness
values. The link is also adjustable in length thus
allowing for variations in inertia. This system is
similar in nature to the control problems
encountered in large geared robot joints where
flexibility is exhibited in the gearbox.
A rigid beam is mounted on a flexible joint
that rotates via a DC motor. The joint deflection
is measured using a sensor. The rotary flexible
joint is an ideal experiment intended to model a
flexible joint on a robot or spacecraft. This
experiment is also useful in the study of
vibration analysis and resonance.
2. Features of ROTFLEX.
High Quality Aluminum chassis with
precision crafted parts
· High Resolution Encoders to sense arm
position
· Variable Loads and Spring Anchors
· Variable Spring Stiffness
· Fully documented system models &
parameters
· Open Architecture Facilitates Matlab /
Simulink
Design
· Modular Design
· 112 Distinct Configurations!
3. Design & Simulation
The rotary flexible joint is an ideal
experiment intended to model a flexible joint on
a robot or spacecraft. This experiment is also
useful in the study of vibration analysis and
resonance. The purpose is to design a state-
feedback controller that will place the tip of the
arm (θ + α) at a given command. Here
requirement is to design a full state feedback
controller using the LQR method to calculate
the gains.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-2
The first part of this will be to design a state-
feedback controller that will meet the required
specifications. The method of calculating the
feedback gains will be the LQR function in
MATLAB's control systems toolbox. The
MATLAB LQR function returns a set of
calculated gains based on the system matrices A
& B and the design matrices Q & R. In this
section begin the iterative design process by
varying Q & R and taking notice on the effect
those changes on the simulated system response
and the closed-loop poles (eigen values).Fix the
Q matrix to be only diagonal. This will allow
you to vary 4 parameters for Q (q1,q2,q3,q4)
and one parameter for R (r1) (R in this case is
scalar as there is only 1 input and therefore 1
control signal).
Q = diag ([10 100 1 1]) & R = 1
This model is a simulation of the Flexible
Joint system with a feedback law u = -kx. The
gain vector k is set by the LQR function. The
largest element of Q should be the one that is
associated with α.Notice that the largest value in
Q is q2. The default Q & R values are
Q = diag ([10 100 1 1]) & R = 1.
4. Basic controllers
An automatic controller compares the actual
values of plant output with the reference input
through feedback elements and produces an
actuating signal, that actuates the control
elements thus reducing the deviation to zero as
to a small valued. The basic controllers
depending upon the manner in which the control
signal is produced.
5. Linear Quadratic Regulator (LQR)
A most effective and widely used technique of
linear control systems design is the optimal
linear quadratic regulator (LQR). The simplified
version of the LQR problem is to find the
control such that the performance index is
minimized for the system.
The formulation of the problem follows.
Given the linear system .
x Ax Bu
y Cx
Find a control function u(t) that will minimize
the cost function J given by
0
1( ' ' )
2J x Qx u Ru dt
The function inside the integral is a quadratic
form and the matrices Q and R are usually
symmetric. It is assumed that R is positive
definite. (i.e. it is symmetric and has positive
eigen values) and Q is positive semi
definite(i.e. it is symmetric and its eigenvalues
are nonnegative).These assumptions imply that
the cost is nonnegative, so its minimum value is
zero. For the cost function to achieve its
minimum value both x and u must go to zero.
This type of control problem is called a
regulator problem. When the state vector is to
track nonzero values, J can be redefined to
create an optimal servo mechanism (tracking)
problem. Regulator behavior is important for
control systems of many types (e.g., attitude
control of satellites or spacecraft, where a zero
reference should be maintained in spite of
disturbances).
6. Properties of the LQR design
LQR has many desirable properties. Among
them are good stability margins and sensitivity
properties. Most of these properties can be
derived using the return-difference inequality
derived by Kalman.
6.1Industrial Applications
Multivariable design based on LQR theory
and the Kalman filter accounts for thousands of
real-world applications.
Geostationary satellite tracking
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-3
Zinc Coating-Mass estimation in
continuous galvanizing lines
Roll-Eccentricity estimation in thickness
control for rolling mills
Vibration control in flexible structures
7. Fuzzy Logic Controller
The tasks of a controller are more complex
than explained earlier for majority of the
systems. In such situations, a fuzzy controller
can be developed to control the system. Fuzzy
logic control is proved to be effective for
complex, non-linear and imprecisely defined
process for which standard model based control
techniques are impractical.
In most of the systems the changes in
parameters due to nose and major disturbances
do occur. In such cases the performance of PID
controller is not that good. Better control
techniques are needed to improve the system
performance.
As fuzzy logic controller is strongly based on
the concepts of fuzzy sets and relations and
linguistic variables, the advantages of fuzzy
logic includes quicker development cycles and
superior performance. Basically, conventional
control systems have five development steps:
developing a mathematical model of system,
developing a mathematical model for a
controller, analyzing the equations, converting
the model to a circuit and building the circuit.
Fuzzy logic, on the other hand has just three:
developing a fuzzy model, simulation and
compilation.
8. Simulations & Results
8.1 Simulation Of ROTFLEX With LQR
Procedure for doing the experiment
Before starting the experiment check the
following:
a. The arm deflection signal (α) should be
connected to encoder channel #1 and the
servomotor's position signal (θ) should be
connected to encoder channel #0.
b. Analog Output channel #0 should be
connected to the UPM (Amplifier) and from the
amplifier to the input of the servomotor. This
system has one input (Vm) and two outputs (θ
& α).
1. Click on start then click on wincon server.
2. Click on start then click on wincon client.
3. Click on start then click on MATLAB and
run the setup
file. This MATLAB script file will setup all the
specific
System parameters and will set the system
state-space
Matrices A, B, C & D.The MATLAB LQR
function
returns a set of calculated gains based on the
system
matrices A & B and the design matrices Q &
R.
4. Now run the file by click on the “run” in the
command window or pressing F5.Then
calculated gains are displayed in the
MATLAB window.
5. Open simulink in matlab window, then run
the simulink file. This model is a simulation of
the Flexible Joint system with a feedback law u
= -kx.The gain vector k is set by the LQR
function. This model has the I/O connection
blocks linking to the physical plant as well as a
simulated block to compare real and simulated
results
6. Then in the simulink model (fig 1.1) set the
amplitude as 20 by double clicking on it.
7. Fig 1.1 above depicts the Rotary Flexible
Joint controller developed for this experiment.
Notice that both the actual system and an exact
simulation are running in parallel thus allowing
comparing the actual and simulated results.
Before “Building” this controller, make sure
that the state-feedback gain k is set according to
final design parameters. Now proceed to
“Build” the controller through the WinCon
menu. After the code has compiled, start the
controller through WinCon and open up two
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-4
Scopes, for theta (measured and simulated
together). Then measure the rise time, steady
state error of measured and simulated theta
Fig 1. 1Circuit diagram of Flexible Joint Controller with LQR
Fig1.2 Simulated response of simulated theta with LQR
Fig 1.3 Simulated response of measured theta with LQR
9. Simulation Of ROTFLEX With Fuzzy Logic Controller.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-5
Fig 1.1 is modified as fig 1.4 to control with fuzzy logic controller.
Fig 1.4 Circuit diagram of Flexible Joint Controller using fuzzy logic controller
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-6
9.1 Procedure for doing the experiment
Before starting the experiment check the
following:
a. The arm deflection signal (α) should be
connected to encoder channel #1 and the
servomotor's position signal (θ) should be
connected to encoder channel #0.
b. Analog Output channel #0 should be connected
to the UPM (Amplifier) and from the amplifier to
the input of the servomotor. This system has one
input (Vm) and two outputs (θ & α).
1. Click on start then click on wincon server.
2. Click on start then click on wincon client.
3. Click on start then click on MATLAB then in
MATLAB window run the setup file. This
MATLAB script file will setup all the specific
system parameters and will set the system state-
space matrices A, B, C & D. The MATLAB LQR
function returns a set of calculated gains based on
the system matrices A & B and the design
matrices Q & R.
4. Further in the script file, there is a section that
will set User-defined controller specification
section set Student_Config = 'MANUAL’. Save
the file, run it. Check whether the file runs or not
by checking whether design
Specifications file appear in the workspace
directory of mat lab window .if appears, Then
minimize it.
5. Click on start in the mat lab window on the
bottom left corner of computer screen, then
toolboxes, then in fuzzy logic click on the FIS
editor.In FIS editor window click on New then on
Mandani model.
6. In FIS editor double click on “input”, then a
membership editor will appears as below .In this
click on edit and mention how many membership
functions are present. Here there are 11(later 15)
membership functions
and for each membership functions define a name.
Then close it.
7.In FIS editor double click on “output”, then a
membership editor will appears as below In this
click on edit and mention how many membership
functions are present. Here there are 11 (later 15)
membership functions and for each membership
functions define a name. Then close it.
NOTE: here first 11 rules then 15 rules are used
to observe how FLC varies for different rules and
to improve the simulation results.
In FIS editor double click on “mandani”, then a
rule editor will appears as below In this rule editor
edit the rules as required. Here first there are 11
rules (then 15 rules for better results), Then save
and close the FIS editor by giving a file name.
Then go to file in the FIS editor import the .FIS
file from disk and export the .FIS file to
workspace. In mat lab window, in workspace
directory check whether .FIS named file appears
or not. If the file appears then experimental
procedure is in correct way.
Then run simulink model .This model has the I/O
connection blocks linking to the physical plant as
well as a simulated block to compare real and
simulated results
Then in the simulink model (fig 1.4) set the
amplitude as 20 by double clicking on it. For gains
before and after fuzzy logic controller with
viewer, set gain=0.04 and gain=13.Double click
on fuzzy logic controller with rule viewer in fig
1.4 .Here type the FIS editor file name, and then
click on ok. Now proceed to “Build” the controller through
the WinCon menu. After the code has compiled, start the
controller through WinCon and open up two Scopes, for
theta (measured and simulated together).Then measure the
rise time, steady state error of measured theta and the values
are tabulated.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-7
Fig 1.5 Simulated response of simulated theta
Fig 1.6 Simulated response of measured theta (a) 11 rules in rule editor
Fig 1.7 Simulated response of measured theta
(b) 15 rules in rule editor
10 Conclusion
The following table is the comparison between
LQR and Fuzzy logic controller.
Controller Rise time
Steady state error
Linear Quadratic
Regulator (LQR) 770 ms - 11 %
Fuzzy logic
controller 240 ms + 13 %
The comparison of performance of real time
system with simulation is done by simultaneously
observing measured and simulated angular
position of flexible joint. It is found that the
response is improved with fuzzy controller as
compared to LQR.
11 References
[1]. Fuzzy control -Kevin M. passion Stephen
yurkovich.
[2]. Fuzzy Logic (intelligence control &
information) -John Yen & Reza Langari.
[3]. Automatic control systems - Kuo
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0203-8
[4]. Control system Engineering – I.J.Nagrath -
M.Gopal
[5]. Modern control Engineering – Katsuhiko
ogata
6[]. Control system design -Graham C.Goodwin,
Stefan F.Graebe -Mario
E.Salgado-Pearson education
[7]. Design of feedback control system 4th edition
Oxford Indian edition Stefani-Shahian -
Savant –Hostetter
[8]. SRV02-(E; EHR) (T)-Rotary Servo Plant
(Quanser) - User Manual
[9]. SRV02-Series-ROTFLEX – Rotary Flexible
Joint (Quanser)- User Manual
[10]. SRV02-Series-Rotary Experiment # 4-
Flexible Joint (Quanser)-Student Handout.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0204-1
Fuzzy Logic Control for a Speed Control of
Induction Motor using Pulse Width Modulation
Mr. Jay Kumar Pandey
Shri Ramswaroop Memorial College of Engineering and Management , Lucknow , U.P
Email id : [email protected]
Abstract
This paper presents design and implements a voltage
source inverter type space vector pulse width modulation
(PWM) for control a speed of induction motor. This
scheme leads to be able to adjust the speed of the motor
by control the frequency and amplitude of the stator
voltage, the ratio of stator voltage to frequency should be
kept constant. The fuzzy logic controller is also
introduced to the system for keeping the motor speed to be
constant
When the load varies. The experimental results in testing
the 0.22 kW induction motor from no-load condition to
rated condition show the effectiveness of the proposed
control scheme. Keywords—Fuzzy logic control, space vector pulse
width modulation, induction motor.
1. Introduction
The Pulse Width Modulation (PWM) method is an
advanced, computation-intensive PWM
method and possibly the best among all the PWM
techniques for variable frequency drive application.
Because of its superior performance characteristics, it has
been finding widespread application in recent year. The
PWM methods discussed so far have only considered
implementation on half bridges operated independently,
giving satisfactory PWM performance. With a machine
load, the load neutral is normally isolated, which causes
interaction among the phases.
This interaction was not considered before in the PWM
discussion [1]-[4]. Recently, Fuzzy logic control has
found many applications in the past decade. This is so
largely because fuzzy logic control has the capability to
control nonlinear, uncertain systems even in the case
where no mathematical model is available for the
controlled system. However, there is no systematic
method for designing and tuning the fuzzy logic
controller.
This means that if the a reliable expert knowledge is not
available or if the controlled system is too complex to
derive the required decision rules, development of a fuzzy
logic controller become time consuming and tedious or
sometimes impossible. In the case that the expert
knowledge is available, fine-tuning of the controller might
be time consuming as well. Furthermore, an optimal fuzzy
logic controller can not be achieved by trial-and-error.
These drawbacks have limited the application of fuzzy
logic control. Some efforts have been made to solve these
problems and simplify the task of tuning parameters and
developing rules for the controller. These approaches
mainly use adaptation or learning techniques drawn from
artificial intelligence or neural network theories.
Application of fuzzy logic control for the control a speed
induction motor using space vector pulse width
modulation is quite new [5].
This paper presents design and implements a voltage
source inverter type space vector pulse width modulation
for control a speed of induction motor. The paper also
introduces a fuzzy logic controller to the SVPWM in
order to keep the speed of the motor to be constant when
the load varies. The speed motor control system is set up
for testing.
The aim of this paper is two-fold. The first is shown the
dynamics response of speed with design the fuzzy logic
controller to control a speed of motor for keeping the
motor speed to be constant when the load varies. The
second aim is shown the phase voltage and line current
waveforms.
2. Inverter for ac drives
2.1 Space Vector Pulse Width Modulation
The SVPWM method considers this interaction of the
phase and optimizes the harmonic content of the three
phase isolated neutral load as shown in Fig. 1.
Fig. 1 Voltage source inverter type 3 phase
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0204-2
The three phase sinusoidal and balance voltages
given by the equations as follows:
Are applied to the three phase induction motor,
using Eq. (4). A three phase bridge inverter, From Fig. 1,
has 8 permissible switching states. Table I gives summary
of the switching states and the corresponding phase-to-
neutral
voltage of isolated neutral machine.
2.2 Simulink Implementation
To implement the algorithm in Simulink, we shall first
assume that the three-phase voltages at the stator
terminals must have the following from Eqs. (1)-(3), the
frequency f and the amplitude m V are variables.
However, theV f control algorithm implies that there is a
relationship between the amplitude of the voltage and the
frequency, i.e. the ration between the two quantities is
constant [2],[3].
3. Design Of A Fuzzy Logic Controller
In drive operation, the speed can be controlled indirectly
by controlling the torque which, for the normal operating
region, is directly proportional to the voltage to
frequency.
The speed is controlled by fuzzy logic controller whose
output is the reference current of the inner dc current
controller. The torque is controlled by varying the dc
current. The drive performance of SVPWM is improved
by employing 2 sets of fuzzy logic controllers. From Fig.
4, one set of fuzzy logic controller is used in the inner
loop for controlling the torque of the motor which is
proportional to DC link current Idc, and another set is
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0204-3
used in the outer loop for controlling the actual motor
speed .
Therefore, the fuzzy logic controllers in the paper will
result the higher accuracy in controlling the v/f/F. A fuzzy
logic controller is proposed to control the speed of the
motor to be constant when the load varies. The speed
error e(k) and the change of speed error ce(k) are
processed through the fuzzy logic controller whose output
is the voltage command * dc I . The current error is
usually processed by current regulator to produce a
control frequency *e ù . This control frequency adjusts the
v/f of SVPWM such that the desired speed of the motor
can be obtained. In the design of a fuzzy logic controller,
seven membership functions were used for both error and
change of error. Membership functions were constructed
to represent the input and output value. The fuzzy logic
controller consists of three stages: fuzzification, control
rules evaluation and defuzzification. Consider the fuzzy
speed control system , where the input signal are e and ce
and the output signal is du, as explained before. Fig. 5
shows the fuzzy sets and corresponding triangular MF
description of each signal. The fuzzy sets are as follows :
Z = Zero, PB = Positive Big, NB = Negative Big, PS =
Positive Small, NS = Negative Small, NVS = Negative
Very Small, PM = Positive Medium NM = Negative
Medium, PVS = Positive Very Small [6].
The universe of discourse of all the variables , covering
the whole region , is expressed in per unit values. All the
MFs are asymmetrical because near the origin, the signals
require more precision. There are seven MFs for e(pu)
and ce(pu) signal , whereas there are nine MFs for the
output. All the MFs are symmetrical for positive and
negative values of the variables. Fig. 6 shows the
corresponding rule table for the speed controller. The top
row and left column of the matrix indicate the fuzzy sets
of the variables e and cue, respectively , and the MFs of
the output variable du(pu) are shown in the body of the
matrix. There may be 7*7 = 49 possible rules in the
matrix , where a typical rule reads as: IF e(pu) is PS AND
ce(pu) is NM THEN du(pu) is NS. Some blocks in the
rule table may remain vacant giving less number of rules.
4. Experimental Setup
The experimental set-up, illustrated in Fig. 7,
implemented to a three phase induction motor which has
the detail as follows: 0.22 kW, 230/400V, 1.03/0.59 A,
50Hz, P.F 0.8 lag and 1410 rpm. The speed of motor
ranging from 0 to 1500 rpm is measured by incremental
encoder 3600 pulse/rev. All current and voltage are
measured using LEM sensors, and both of them are then
transformed to be a voltage ranging from 0 to 10 volts
which will be the input of A/D respectively. This scheme
enables the user to adjust the speed of the motor by the
duty cycle of the V/F operating in SVPWM mode. The
performances of a linear control technique implemented
on a fuzzy logic controller to control speed of motor using
dSPACE the real-time DS1104’TMS3204 DSP Controller
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0204-4
Board along with the Matlab/Simulink tool with sampling
time 1 ms as shown in Fig. 7 [7]-[11].
4.1 Step Response of Speed
To evaluate the performance of the system, a series of
measurements has been accomplished. The measurements
can be divided in two groups: the first is a step change of
the speed reference at constant load torque and the second
is a step change of the load torque at constant speed
reference. Figs. 8- 9 as shown performance of the fuzzy
logic controller with a fuzzy tuning rule based on step
response of speed control. To be tested time response of
speed, duty cycle and line current via the step change of
speed reference 300 to 1200 rpm with the load torque
equal to zero and equal to rated respectively. Figs. 10-11
as shown time response of speed, duty cycle and line
current via the step change of the load torque at constant
speed reference 600 and 1200 rpm respectively.
From the results tested the performance of
controller by a step change of the speed reference at
constant load torque as shown in Figs. 8-9, it’s found that
the Rise time 2 r p
Maximum overshoot 12 p
load torque at constant speed reference as shown in Figs.
10-11, it’s found that the Settling time 15 s
experimental results obtained, the proposed fuzzy logic
controller can keep the motor speed to be constant at the
speed ranging from 300 to 1200 rpm. Figs. 12-14 as
shown steady state error of speed at reference speed 300,
600 and 1200 rpm rated load respectively. It’s found that
it have state error ±10 rpm.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0204-5
4.2. The Phase Voltage and Input Line Current
Waveforms
The line voltage and line current are measured;
they are measured by using LEM sensors with ratio
Amp/volt and ratio 100 V/volt. All data of signal are kept
on of digital storage oscilloscope. The waveforms of
SVPWM, phase voltage an V and line current a I are
measured digital storage oscilloscope as shown in Figs.
15-20.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0204-6
5. Conclusion
The Fuzzy logic controller is applied to speed signal
model of motor and is then downloaded to dSPACE
through Simulink. The experimental results are analyzed
in testing the 0.22 kW induction motor from zero load
condition to rated condition, it’s found that the speed of
the induction motor can be controlled. In addition, the
motor speed to be constant when the load varies.
References
[1] Andrzej M. Trzynadlowski, Introduction to Modern Power
Electronics, Copyright© 1998 by John Wiley & Sons, Inc. All
rights reserced.
[2] Bimal K.Bose, Modern Power Electronics and AC Drives, ©
2002 Prentice Hall PTR.
[3] W. Leonharn, Control of Electrical Drives, Springer-Verlag
Berlin Heidelberg, New York, Tokyo, 1985.
[4] F. Ashrafzadeht, E.P. Nowickit, and J.C. Samont, “A Self-
Organizing and Self-Tuning Fuzzy Logic Controller for Field
Oriented Control of Induction Motor Drives”, 0-7803-3008-0/95
$4.00© 1995 IEEE, pp. 1656-1662.
[5] Zdenko Kovaccic and Stjepan Bogdan, Fuzzy Controller
design Theory
and Applications, © 2006 by Taylor & Francis Group.
International, 2002.
[6] Hoang Le-Huy and Maher Hamdi, “Control of Direct-Drive
DC Motor by Fuzzy Logic”, 0-7803-1462-x/93$03.00 © 1993
IEEE, pp. 732-738.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-1
Optimal Placement Techniques of FACTS Controllers
in Multi-Machine Power System Environments: A
Literature Survey
Bindeshwar Singh, N. K. Sharma and A. N. Tiwari
Abstract- This paper presents exhaustive review of various
methods/techniques for optimal choice and allocation of
FACTS controllers. Authors strongly believe that this
survey article will be very much useful to the researchers
for finding out the relevant references in the field of
placement of FACTS Controllers.
Index Terms- Flexible AC Transmission System (FACTS),
FACTS Controllers, Placements, Voltage Stability.
I.INTRODUCTION
HE drive towards deregulated environment may result in
simultaneous installation of different FACTS controllers
in power system. These multiple FACTS controllers have the
potential to interact with each other. This interaction may
either deteriorate or enhance system stability depending upon
the chosen controls and placement of FACTS controllers.
Hence there is a need to study the interaction between the
FACTS controllers.
The various interactions can potentially occur between the
different FACTS controllers, as well as, between FACTS
controllers and Power System Stabilizers (PSS) in a multi-
machine power system environment. These likely interactions
have been classified into different frequency ranges and
various interaction problems between FACTS controllers or
FACTS to PSS’s from voltage stability/ small signal stability
viewpoint are presented in [1]-[2].
There are several methods proposed in literatures [3]-[86],
[87]-[111], for placement of FACTS controllers in multi-
machine power systems from different operating conditions
viewpoint. References [3]-[5], classify three broad categories
such as a sensitivity based methods, optimization based
method, and artificial intelligence based techniques for
placement of FACTS controllers from different operating
conditions viewpoint in multi-machine power systems. The
various sensitivity based methods have been proposed in
literatures includes eigen-value analysis based methods [6]-
[12], modal analysis techniques [13]-[15], index methods [16]-
[23], residue-based methods [24]-[25],[43],[98]-[99], and
other sensitivity based methods [26]-[37],[53],[87]-[97],[111].
The various optimization based methods have been proposed
in literatures that includes non-linear optimization
programming techniques [38],[39],[103], mixed integer-
optimization programming techniques [40]-[42],[100]-[101],
dynamic optimization programming algorithms [44], hybrid
optimization programming algorithms [45], bellmann’s
optimization principle [46], decomposition coordination
methods [47]-[48], curved space optimization techniques
[104]. The various artificial intelligence (AI) based methods
proposed in literature includes genetic algorithms (GA) [49]-
[64],[105]-[106], [110], tabu search algorithms [65],[66],
simulated annealing (SA) based approach [69]-[70],[107],
particle swarm optimization (PSO) techniques [71]-[73],[80],
artificial neural network (ANN) based algorithms [74]-[76],
ant colony optimization (ACO) algorithms [77]-[78], graph
search algorithms [79], fuzzy logic based approach [81]-[82],
other techniques such as norm forms of diffeomorphism
techniques [83], evolution strategies algorithms [84],[86],
improved evolutionary programming [68], gravitational
optimization techniques [85], benders decomposition
techniques [42], augmented Lagrange multiplier approach
[67], hybrid meta-heuristic approach [102], heuristic and
algorithmic approach [108], energy approach [109].
This paper is organized as follows: Section 2 presents the
review of various techniques/methods for placements of
FACTS controllers in multi-machine power systems. Section 3
presents the summary of the paper. Section 4 presents the
conclusions of the paper.
II. CLASSIFICATIONOF FACTS CONTROLLERS
ALLOCATION TECHNIQUES
Three broad categories of allocation techniques for
determining best suited location of FACTS controllers are
sensitivity based methods, optimization based method, and
artificial intelligence based techniques [3]-[5].
A. Sensitivity Based Methods
There are various sensitivity based methods such as a modal
or eigen-value and index analysis. An eigen-value analysis
based techniques has been proposed in [6] for the selection of
parameters of stabilizers in multi-machine power system to
enhance the damping of the power system oscillations.
Reference [7], presents an eigen-value analysis based
algorithms such as a participation factor method for
identification of optimum location of power system stabilizers
to enhance the damping of power system oscillations. Many
literatures shows the existing methods of identifying the
optimum sites for installing PSSs in multi-machine power
systems are restricted to the sequential PSS application, which
T
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-2
considers the enhancement of damping of just one critical
electro-mechanical modes at a time and the eigen structure
analysis of the open loop system, which does not take the
control matrix (the B matrix in the linearized model dX/dt=
AX+BU for power system) into considerations. An eigen-
value analysis based techniques has been proposed for
identifying the optimum sites for installing power system
stabilizers (PSSs) in multi-machine power systems in [8]. The
advantages of this proposed method is that it can identify the
optimum sites for installing PSS so that several electro-
mechanical modes are damped out simultaneously and it takes
both the eigen structure of the open loop system and the
control matrix into consideration. Reference [9], suggests an
eigen-value analysis based approach for identifying the most
effective FACTS controllers, locations, types and ratings that
increase asset utilization of power systems. Reference [10],
uses participation factor method has been suggested for the
critical mode are used to determine the most suitable sites for
SVC (Static Var Compensator) for system reinforcement. In
[11]-[12], an eigen-value analysis based approach has been
proposed for find the optimal location and rating of FACTS
controllers (Static Var Compensator (SVC) and Thyristor
Controlled Series Controller(TCSC)) and a continuation
power flow is used to evaluate the effects of SVC and TCSC
devices on power loadability). In [13] , a modal analysis
algorithm has been suggested for allocation and control of
FACTS devices for steady-state stability enhancement of large
scale power system. A modal analysis algorithm has been
suggested for placement of SVCs and selection of stability
signals in power systems environments [14]. A new eigen
solution free method of modal control analysis for the
selection of the robust installing locations and feedback
signals of FACTS based stabilizers in large–scale power
systems is presented in [15]. An index based approach known
as a Location Index for Effective Damping (LIED) method has
been proposed for identifying the location of SVC and a
Variable Series Capacitor (VSrC) in large-scale power
systems for damping effectively in [16]. A structure
preserving energy margin sensitivity based analysis has been
addressed for determine the effectiveness of FACTS devices
to improve transient stability of a power system in [17]. A
controllability index method has been proposed for select the
input signals for both single and multiple FACTS devices in
small and large power systems for damping inter-area
oscillations in [18]. Different input output controllability
analysis are used to asses the most appropriate input signals
(stability signals) for the SVC, the Static Synchronous Series
Compensator (SSSC) and the Unified Power Flow Controller
(UPFC) for achieving good damping of inter-area oscillations.
A new method called the Extended Voltage Phasors Approach
(EVPA) has been suggested for placement of FACTS
controllers in power systems for identifying the most critical
segments/bus in power system from the voltage stability view
point in [19]. In [20]-[23], an index based method has been
addressed for determine the suitable locations of FACTS
devices in power system environments. A residues based
approach has been proposed for allocation of FACTS
controllers in power system to enhance the system stability
[24]-[25]. An efficient algorithm for the solution of two
important problems in the area of damping control of electro-
mechanical oscillations in large scale power systems has been
proposed in [26]. The proposed algorithms allow the
determination the most suitable generators for installing power
system stabilizers and the most suitable buses in the system
for placing SVC in order to damp the critical modes of
oscillation. A sensitivity based approach has been proposed
for placement of FACTS controllers in open power markets to
reduce the flows in heavily loaded lines, resulting in an
increased loadability, low system loss, improved stability of
the network, reduced cost of production and fulfilled
contractual requirement by controlling the power flows in the
network in [27]-[28] . A sensitivity based approach called Bus
Static Participation Factor (BSPF) has been proposed for
determine the optimal location of static VAR compensator
(SVC) for voltage security enhancement in [29]. A sensitivity
based approach has been proposed to determine the placement
of TCSC and UPFC for enhancing the power system
loadability [30]. In [31], a sensitivity analysis method has been
proposed for determine the optimal placement of static VAR
compensator (SVC) for voltage security enhancement in
Algerian Distribution System. In [32], a sensitivity analysis
and evolutionary programming techniques has been proposed
for determine the optimal placement of UPFC in power system
from the operational planning viewpoint. Sensitivity analysis
is superior when compared to the others as sensitivity analysis
gives the best possible installation location. References [33]-
[37], presents a sensitivity based approach has been proposed
for optimal placement of UPFC to enhance voltage stability
margin under contingencies. Reference [87], suggests a
normal form analysis approach based on sensitivity indices is
used for sitting power systems stabilizers (PSS) in power
systems network. Reference [88], suggested a Trajectory
Sensitivity Analysis (TSA) technique for the evaluation of the
effect of TCSC placement on transient stability. A sensitivity
based technique is used for placement of Static Synchronous
Series Compensator (SSSC) in power system network in [89].
Sensitivity based screening technique for greatly reducing the
computation involved in determining the optimal location of a
Unified Power Flow Controller (UPFC) in a power system
[90]. A sensitivity analysis and evolution programming
technique has been proposed for placement of UPFC in a
power system in [91]. In [92], a sensitivity based technique
used for determine the minimum amount of shunt reactive
power (VAr) support which indirectly maximizes the real
power transfer before voltage collapse is encountered.
Sensitivity information that identifies weak buses is also
available for locating effective VAr injection sites. A
sensitivity based technique is used for determine optimal
placement of Static Synchronous Compensator (STATCOM)
and Unified Power Flow Controllers (UPFC) to enhancement
of Dynamic Available Transfer Capability (ATC) under
different contingencies in New England System [93]. A
sensitivity factor based approach has been used in [94] for the
optimal placement of the TCSC to minimize the congestion
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-3
cost. In [95], a second-order sensitivity analysis technique
used for optimal location of SVC and TCSC in power system.
An eigen-value sensitivity based approach has been used for
location and controller design of TCSC to enhance damping
power system oscillations [96]. In [97], a new sensitivity
factor, called System Loading Distribution Factor is used for
determine the optimal location of UPFC in power system. A
residues factor method has been used for determination of
optimal location of the TCSC device to damp out the inter-
area mode of oscillations [98]. In [99], a residues method
based on sensitivity analysis technique is used for determine
optimal location of the SVC to enhance the damping power
system oscillations. A new approach based on sensitivity
indices has been used for the optimal placement of various
types of FACTS controllers such as TCSC, TCPAR and SVC
in order to minimize total system reactive power loss and
hence maximizing the static voltage stability in [111]. Magaji
and Mustafa et al. [43] has been suggested a residue factor
approach for optimal location of FACTS devices for damping
oscillations of power systems. S. N. Singh and I. Erlich et al.
[53] proposed for locating UPFC for enhancing power System
loadability.
B. Optimization Based Techniques
This section reviews the optimal placement of FACTS
controllers based on various optimization techniques such as a
linear and quadratic programming, non-linear optimization
programming, integer and mixed integer optimization
programming, and dynamic optimization programming.
1. Non-Linear Optimization Programming (NLP) techniques
When the objective function and the constraints are non-linear,
it forms non-linear programming (NLP). The difference
between NLP and Linear Programming (LP) is analogous to
the difference between a set of solving non-linear equations
and a set of solving linear equations. In most of the NLP
methods, the approach is to start from an initial guess and to
determine a descent direction in which objective function
decreases in case of minimization problems [5].
Reference [38], suggests a non-linear optimization
programming techniques for assessing the placement of
FACTS controllers in power system to damp electro-
mechanical oscillations. A non-linear optimization
programming techniques has been proposed for optimal
network placement of SVC controller in [39] and a Benders
Decomposition technique has been used for these solutions.
2. Integer and Mixed –Integer optimization Programming
(IP & MIP) techniques
Reference [40], a mixed integer linear optimization
programming algorithm has been proposed for the optimal
placement of TCPST in multi-machine power systems. A
mixed integer optimization programming algorithm has been
proposed for optimal placement of Thyristor Controlled Phase
Shifter Transformers (TCPSTs) in large scale power system
for active flow and generation limits, and phase shifter
constraints in [41]. A mixed integer optimization
programming algorithm has been proposed for allocation of
FACTS controllers in power system for security enhancement
against voltage collapse and corrective controls, where the
control effects by the devices to be installed are evaluated
together with the other controls such as load shedding in
contingencies to compute an optimal VAR planning [42]. In
[100], a mixed integer non-linear optimization programming
algorithm is used for determine the type, optimal number,
optimal location of the TCSC for loadability enhancement in
deregulated electricity markets. A mixed integer optimization
programming algorithm has been used for optimal location of
TCSC in a power system in [101].
3. Dynamic Programming (DP) techniques
Oliveira et al. suggested a dynamic optimization programming
algorithm for allocation of FACTS devices in hydrothermal
systems in order to minimize the expected thermal generation
costs and the investments on FACTS devices in a pre-
specified time interval [44]. Chang and Huang et al. showed
that a hybrid optimization programming algorithm for optimal
placement of SVC for voltage stability reinforcement [45]. In
[46], a bellmann’s optimality principle has been proposed for
optimal sectionalizing switches allocation in distribution
networks and also determines the optimal number of
automatic sectionalizing switches devices. Lie and Deng et al.
has addressed a decomposition coordination technique for
optimal FACTS devices allocation in power system economic
dispatch [47]. Zuwei and Lusan et al. presented review on the
current status on the optimal placements of FACTS devices in
power systems for enhances the damping of power system
oscillations [48]. Orfanogianni and Bacher et al. suggested an
optimization-based methodology is used for identify key
locations of TCSC and UPFC include the nonlinear constraints
of voltage limitation, zero megawatt active power exchange,
voltage control, and reactive power exchange in the ac
networks [103]. In [104], Curved Space Optimization (CSO)
programming algorithm is used for allocation of SVC in a
large power system.
C. Artificial Intelligence (AI) Based Techniques
This section reviews the optimal placement of FACTS
controllers based on various Artificial Intelligence based
techniques such as a Genetic Algorithm (GA), Expert System
(ES), Artificial Neural Network (ANN), Tabu Search
Optimization (TSO), Ant Colony Optimization (ACO)
algorithm, Simulated Annealing (SA) approach, Particle
Swarm Optimization (PSO) algorithm and Fuzzy Logic based
approach.
1. Genetic Algorithm(GA)
A genetic algorithm has been addressed for optimal location of
phase shifters in the French network to reduce the flows in
heavily loaded lines, resulting in an increased loadability of
the network and a reduced cost of production [49]. A genetic
algorithm has been addressed for optimal location of multiple
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-4
type FACTS controllers in a power system. The optimization
are performed on three parameters; the location of the devices,
their types and their values. The system loadability is applied
as measure of power system performance. Four different kinds
of FACTS controllers are used as models for steady state
studies: TCSC, TCPST, Thyristor Controlled Voltage
Regulator (TCVR) and SVC in order to minimizing the overall
system cost, which comprises of generation cost and
investment cost of FACTS controllers [50]. Vijakumar and
Kumudinidevi et al. presented a novel method for optimal
location of FACTS controllers in a multi-machine power
system. The location of FACTS controllers, their type and
rated values are optimized simultaneously for the various
FACTS controllers, TCSC and UPFC are considered in order
to minimizing the overall system cost, which comprises of
generation cost and investment cost of FACTS controllers
[51]. A stochastic searching algorithm called as genetic
algorithm has been proposed for optimal placement of static
VAR compensator for enhancing voltage stability in [52]. In
[54], a genetic algorithm (GA) based method has been
proposed for determine the optimal placement of FACTS
controllers in power system with the consideration of
economics and cost effectiveness. In [55], a genetic algorithm
(GA) based approach has been proposed for the optimal
choice and allocation of FACTS devices in deregulated
electricity power market is to achieve the power system
economic generation allocation and dispatch in deregulated
electricity market. The locations of the FACTS controllers,
their type and ratings are optimized simultaneously. Reference
[56], genetic algorithm (GA) and particle swarm optimization
(PSO) has been proposed for optimal location and parameter
setting of UPFC for enhancing power system security under
single contingencies. A new genetic algorithm (GA) based
approach [57] has been addressed for selection of the optimal
number and location of UPFC devices in deregulated electric
power systems. In [58], a novel method such as a genetic
algorithm has been presented for optimal location of FACTS
controllers in a multi-machine power system. The location of
FACTS controllers, their type and rated values are optimized
simultaneously for the various FACTS controllers such as a
TCSC and UPFC are considered. Reference [59], a genetic
algorithm (GA) has been proposed for location and parameters
setting of UPFC for congestion management in pool market
model. The planning method such as a micro-genetic
algorithm (MGA) has been addressed for optimal type
selection and placement of a FACTS device for power system
stabilizing purpose in a multi-machine power system [60]. A
heuristic approach using genetic algorithm has been addressed
for optimal location of FACTS controllers in multi-machine
power systems for enhancing the damping of power system
oscillations in [61]-[64]. In [105], a non-traditional
optimization technique, a Genetic Algorithm (GA) is
conjunction with Fuzzy logic (FL) is used to optimize the
various process parameters involved in introduction of FACTS
devices such as a TCSC, Thyristor Controlled Phase Angle
Regulator (TCPAR), SVC and UPFC in a power system. The
various parameters taken into consideration were the location
of the device, their type, and their rated value of the devices. A
multi-objective optimal power flow and genetic algorithms
used to determine the optimal number and location of UPFC
devices in an assigned power system network for maximizing
system capabilities, social welfare and to satisfy contractual
requirements in an open market power [106]. An energy
approach has been used for the optimal location of FACTS
controllers/sensors in large-scale power systems in [109].
Reference [110], a genetic algorithm (GA) has been proposed
for optimal choice and allocation of FACTS devices such as
UPFC, TCSC, TCPST, and SVC in deregulated electricity
market.
2. Tabu Search Algorithm (TS)
A TS algorithm has been addressed for optimal placement of
FACTS controllers such as TCSC, TCPST, UPFC, and SVC
in multi-machine power systems [65]-[66]. Reference [102], a
hybrid-meta heuristic method based on tabu search in
conjunction with evolutionary particle swarm optimization
technique has been proposed for optimal location of UPFC in
power system.
3. Simulated Annealing (SA) Algorithms
References [69], [70], a novel optimization based
methodology such as a simulated annealing has been
proposed for optimal location of FACTS devices such as
TCSC and SVC in order to relive congestion in the
transmission line while increasing static security margin and
voltage profile of power system networks. In [107], the Goal
Attainment (GA) method based on the SA approach is applied
to solving general multi-objective VAR planning problems by
assuming that the Decision Maker (DM) has goals for each of
the objective functions. The VAR planning problem involves
the determination of location and sizes of new compensators
considering contingencies and voltage collapse problems in a
power system.
4. Particle Swarm Optimization (PSO) Algorithms
In [71], a Particle Swarm Optimization (PSO) algorithm has
been addressed for the solution of the Optimal Power Flow
(OPF) using controllable FACTS controllers to enhance
economic solution. Rashed et al. suggested a Genetic
Algorithm (GA) and PSO techniques for optimal location and
parameter setting of TCSC to improve the power transfer
capability, reduce active power losses, improve stabilities of
the power network, and decrease the cost of power production
and to fulfill the other control requirements by controlling the
power flow in multi-machine power system network [72]. A
Trinary Particle Swarm Optimization Technique has been
proposed for optimal switch placement in distribution systems
for achieving high distribution reliability levels and con-
currently minimizing capital costs can be considered as the
main issues. A novel three state approaches has been proposed
for inspired from the discrete version of a powerful heuristic
algorithm, PSO is developed and presented to determine the
optimal number and locations of two types of switches
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-5
(sectionalizes and breakers) in radial power systems
automation is an important issue from the reliability and
economical point of view [73]. In [74]-[76], an Artificial
Intelligence Based Techniques has been addressed for optimal
placement of FACTS controllers in large scale power system.
In [80], a Particle Swarm Optimization (PSO) technique has
been addressed for optimal location of FACTS controllers
such as TCSC, SVC, and UPFC considering system
loadability and cost of installation.
5. Ant Colony Optimization (ACO) algorithms
In [77], a methodology has been suggested for placement of
sectionalizing switches in distribution networks in the
presence of distributed generation sources for reliability
improvement with consideration of economic aspects. In [78],
an ACO algorithm has been addressed for the planning
problem of electrical power distribution networks, stated as a
mixed non-linear integer optimization problem, is solved
using the Ant Colony System (ACS) algorithm. The ACS
methodology is coupled with a conventional distribution
system load flow algorithm and adapted to solve the primary
distribution system planning problem. A Graph Search
Algorithm has been addressed for optimal placement of fixed
and switched capacitors on radial distribution systems to
reduce power and energy losses, increases the available
capacity of the feeders, and improves the feeder voltage
profile [79].
6. Fuzzy Logic (FL) Algorithms
References [81]-[82], A fuzzy logic based approach has been
addressed for optimal placement and sizing of FACTS
controllers in power systems. In [83], the theory of the normal
forms of diffeomorphism algorithm has been addressed for the
SVC allocation in multi-machine power system for power
system voltage stability enhancement. Luna and Maldonado et
al. has been addressed a new methodology is based on the
evolutionary strategies algorithm known as Evolution
Strategies (ES) for optimally locating FACTS controllers in a
power system for maximizes the system loadability while
keeping the power system operating within appropriate
security limits [84]. A Gravitational Optimization (GO)
algorithm has been addressed for allocation of SVC in a power
system in [85]. Kalyani et al. [86] has been suggested an
Evolutionary algorithm for optimal location of UPFC and
sequential quadratic programming (SQP) to optimize the
UPFC control settings. In [108], a knowledge and algorithm
based approach is used to VAR planning in a transmission
system. The VAR planning problem involves the
determination of location and sizes of new compensators
considering contingencies and voltage collapse problems in a
power system. Fang and Ngan et al. [67] suggested an
augmented Lagrange Multipliers approach for optimal
location of UPFC in power systems to enhances the steady
state performance and significantly increase the loadability of
the system. Hao et al. [68] has been proposed an improved
evolutionary programming technique for optimal location and
parameters settings of UPFCs to maximize the system
laudability subject to the transmission line capability and
specified voltage level.
III. SUMMARY OF THE PAPER
The following tables give summary of the paper as:
A. Methods/Techniques for Placement of FACTS
controllers
1. Methods/Techniques point of view
Methods/Techniques Total No. of
Literatures Reviews out of
106 Literatures
% of Literatures
Reviews out of 106 Literatures
Sensitivity based methods 48 45.28
Optimization based methods 14 13.20
AI-based techniques 44 41.51
2. Operating Parameters point of view
Operating Parameters of Power
systems Total No. of Literatures
Reviews out of
106 Literatures
% of Literatures Reviews out of
106 Literatures
Damping of power system
oscillations
16 15.09
Voltage Profile 20 18.87
Security of the power system 02 1.89
Small signal stability, transient stability
06 5.66
Power transfer capability
through the lines
02 1.89
Dynamic performances of
power systems
02 1.89
The loadability of the power
system network
12 11.32
Others parameters point of view 46 43.39
From above tables it is concluded that the 45.28% of total
literatures are reviews based on sensitivity methods, 13.20%
of total literatures are reviews based on optimization
programming and the 41.51% of total literatures are reviews
on AI based techniques regarding with placement of FACTS
controllers in multi-machine power systems. It is also
concludes that the 15.09% of total literatures are reviews
carryout from damping of power system oscillations, 18.87%
of total literatures are reviews carryout from voltage stability,
1.89% of total literatures are reviews carryout from security of
power system, 5.66% of total literatures are reviews carryout
from small signal/transient/dynamic stability, 1.89% of total
literatures are reviews carryout from power transfer capability
through the lines, 1.89% of total literatures are reviews
carryout from dynamic performance of power system, 11.32%
of total literatures are reviews carryout from the loadability of
power system, and 43.39% of total literatures are reviews
carryout from other parameters viewpoints.
Finally it is concluded that the maximum research work
carryout from damping of power system oscillations and
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-6
voltage stability point of view regarding with placement of
FACTS controllers in multi-machine power systems.
IV. CONCLUSIONS
In this paper an attempt has been made to reviews, various AI
based optimization methods used for the placement of FACTS
controllers. Even through, excellent advancements have been
made in classical methods i. e. sensitivity based method, they
suffer with the following disadvantages: In most cases,
mathematical formulations have to be simplified to get the
solutions because of the extremely limited capability to solve
real-word large-scale power system problems. They are weak
in handling qualitative constraints. They have poor
convergence, may get stuck at local optimum, they can find
only a single optimized solution in a single simulation run,
they become too slow if number of variables are large and
they are computationally expensive for solution of a large
power system.
Whereas, the major advantage of the AI based optimization
methods is that they are relatively versatile for handling
various qualitative constraints. AI methods can find multiple
optimal solutions in single simulation run. So they are quite
suitable in solving multi-objective optimization problems for
placement FACTS controllers in multi-machine power system.
In most cases, they can find the global optimum solution. The
main advantages of ANN are: possesses learning ability, fast,
appropriate for non-linear modeling, etc. whereas, large
dimensionality and the choice of training methodology are
some disadvantages of ANN. The advantages of Fuzzy
method are: Accurately represents the operational constraints
and fuzzified constraints are softer than traditional constraints.
The advantages of GA methods are: It only uses the values of
the objective function and less likely to get trapped at a local
optimum. Higher computational time is its disadvantage. The
advantages of EP are adaptability to change, ability to
generate good enough solutions and rapid convergence. ACO
and PSO are the latest entry in the field of optimization. The
main advantages of ACO are positive feedback for recovery of
good solutions, distributed computation, which avoids
premature convergence. It has been mainly used in finding the
shortest route in transmission network, short-term generation
scheduling and optimal unit commitment. PSO can be used to
solve complex optimization problems, which are non-linear,
non-differentiable and multi-model. The main merits of PSO
are its fast convergence speed and it can be realized simply for
less parameters need adjusting. PSO has been mainly used to
solve bi-objective generation scheduling, optimal reactive
power dispatch and to minimize total cost of power
generation. The applications of ACO and PSO for placement
of FACTS controllers in multi-machine power system.
This paper has also addressed a survey of several technical
literature concerned with various techniques/methods for
placement FACTS controllers in multi-machine power system
environments to show that the achieve significant
improvements in operating parameters of the power systems
such as small signal stability, transient stability, damping of
power system oscillations, security of the power system, less
active power loss, voltage profile, congestion management,
quality of the power system, efficiency of power system
operations, power transfer capability through the lines,
dynamic performances of power systems, and the loadability
of the power system network also increased. This review also
shows that installing multiple controllers on the system may
not improve the dynamic performance due to undesirable
interactions. The tuning of one controller may affect other
controllers and thus lead to unstable conditions. These issues
should be taken into consideration when designing systems
with multiple controllers. The implementation of a
coordinated controller tuning procedure to avoid undesirable
interactions in power systems, and thus improve overall
dynamic performance is under this review.
Authors strongly believe that this survey article will be very
much useful to the researchers for finding out the relevant
references as well as the previous work done in the field of
placement of FACTS Controllers for the various
methods/techniques for placement of FACTS controllers in
multi-machine power systems. So that further research work
can be carried out.
ACKNOWLEDGMENT
The authors would like to thanks Dr. S. C. Srivastava, and Dr.
S. N. Singh, Indian Institute of Technology, Kanpur, U.P.,
India, and Dr. K.S. Verma, and Dr. Deependra Singh, Kamla
Nehru Institute of Technology, Sultanpur, U.P., India, for their
valuables suggestions regarding placement and coordination
techniques for FACTS controllers form voltage stability, and
voltage security point of view in multi-machine power
systems environments.
REFERENCES
[1] M. J. Gibbard , D. J. Vowles, and P. Pourbeik, “Interactions between
and Effectiveness of Power Systems Stabilizers and FACTS controllers in Multi-Machine Systems,” IEEE Trans. on Power
Systems, Vol. 15, No. 2, pp. 748-755, May 2000.
[2] Kundur P., “Inter-area Oscillations in Power System,” IEEE Power Engineering Society, pp. 13-16, October 1994.
[3] W. Zhang, F. L., and Leon M. Tolbert, “Optimal Allocation of Shunt
Dynamic VAR Source SVC and STATCOM: A Survey”, IEEE Trans on Power systems, Vol.22, No. 3, Oct 2007.
[4] H. N. Ng, M. M. A. Salama, and A. Y. Chikhani, “ Classification of
Capacitor Allocation Techniques” IEEE Trans. on Power Delivery , Vol. 15, No.1, Jan 2000.
[5] R. C. Bansal, “Optimization Methods for Electric Power Systems: An
Overview,” International Journal of Emerging Electric Power Systems, Vol.2, issues-1, 2005
[6] R. J. Fleming, M. A. Mohan, and K. Parvatisam, “Selection of
Parameters of Stabilizers in Multi-machine Power System” IEEE Trans on Power Apparatus and Systems, Vol.PAS-100, No.5, May
1981.
[7] Yuan–Yih Hsu, and Chern-Linchen, “Identification of Optimum Location for Stabilizers Application Using Participation Factors,” IEE
Proc., Vol. 134, Pt.C. No. 3, May 1987.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-7
[8] Jin Lu Hsiao-Dong Chiang and James S. Thorp, “Identification of
Optimum sites for Power System Stabilizers Applications,” IEEE
Trans on Power Systems, Vol.5, No.4, Nov.1990. [9] L. A. S. Pilotto, W. W. Ping, A. R. Carvalino, A. Wey, W. F. Long, F. L.
Alvarado, and A. Edris, “ Determination of Needed FACTS
Controllers that Increase Asset Utilization of Power Systems,” IEEE Trans on Power Delivery, Vol.12, No.1, Jan.1997.
[10] Y. Mansour, W. Xu. F. Alvarado, and C. Rinzin, “SVC Placement Using
Critical Modes of Voltage Stability,” IEEE Trans. on Power Systems, Vol. 9, pp. 757-762, May 1994.
[11] A. Kazemi, and B. Badrzadeh, “Modeling and Simulation of SVC and TCSC to Study Their Limits on Maximum Loadalibility Point,”
Electrical Power & Energy Systems, Vol. 26, pp. 619-626, 2004.
[12] Cigre Working Group, “Modeling of Static VAR Systems for Systems Analysis,” Electra, Vol. 51, pp.45-74, 1977.
[13] Guohong Wu, Akihiko Yokoyama, Jiali He ,and Yixin Yu , “ Allocation
and Control of FACTS controllers for Steady State Stability Enhancement of Large-Scale Power systems,” IEEE Trans on Power
Systems , Vol. 13, No. 2, 1998.
[14] Malihe M. Farsangi, Hossein Mezamabadi-pour, “Placement of SVCs
and Selection of Stabilizing Signals in Power Systems, “IEEE Trans.
on Power Systems, Vol. 22, No. 3, August 2007.
[15] H. F. Wang, “Selection of Robust Installing Locations and Feedback Signals of FACTS Based Stabilizers in Multi-Machine Power
systems,” IEEE Trans on Power Systems, Vol. 14, No.2, May 1999.
[16] H. Okamoto, A. Kurita, and Y. Seking, “A method for Identification of Effective Locations of Variable Impedance Apparatus on
Enhancement of Steady-State Stability in Large Scale Power
Systems,” IEEE Trans on Power Systems, Vol. 10, pp. 1401-1407, August 1995.
[17] K. N. Shubhanga, and Anil M. Kulkarni, “Application of Structure
Preserving Energy Margin Sensitivity to Determine the Effectiveness of Shunt and Series FACTS Devices,” IEEE Trans. on Power
Systems, Vol. 17, No.3, August 2002.
[18] M. M. Farsang, Y. H. Song, and Kwang Y Lee, “Choice of FACTS device Control Inputs for Damping Inter-area Oscillation,” IEEE
Trans. on Power Systems, Vol. 19, No. 2, May 2004.
[19] N. K. Sharma, A. Ghosh, and R. K. Verma, “A Novel Placement Strategy for FACTS Controllers”, IEEE Trans. on Power Delivery,
Vol. 18, No.3, July 2003.
[20] A. Mohamed and G. B. Jasmon, “Determining the weak Segment of Power System with Voltage Stability Considerations,” Electrical
Power & Energy Systems, Vol. 24, pp. 555-568, 1996.
[21] T. T. Lie, “Method of Identifying the Strategic Placement for Compensation devices,” IEEE Trans. on Power Systems, Vol. 10,
No.3, pp. 1448-1453, August 1995.
[22] H. F. Wang , F. J. Swift, and M. Li, “Indices for Selecting the Best Location of PSSs or FACTS Based Stabilizers in Multi-Machine
Power Systems: A Comparative Study ,” IEE Proc., Pt. C, Vol. 144,
No. 2, pp. 155-159, March 1997. [23] B. Kalyan Kumar, S. N. Singh and S. C. Srivastava, “Placement of
FACTS Controllers Using Modal Controllability indices to Damp
out Power System Oscillations,” IET Gener. Transm. Distrib. Vol.1, No.2, March 2007.
[24] O. O. Obadina and G. J. Berg, “Identifying Electrically Week and
Strong Segments of a Power Systems from a Voltage Stability Viewpoint,” IEE Proc. Pt.C, Vol-137, No.3, pp. 205-212, May 1990.
[25] L. A. S. Pilotto, W. W. Ping , A. R. Carvalho, W. F. Long, F. L.
Alvarado, C. L. Demarco, and A. Edris, “Coordinated Design of FACTS Controllers to Enhance Power System Dynamic Performance
,” International Colloquium on HVDC and FACTS , Johannesburg,
Sept. 1997. [26] N. Martins and L. T. G. Lima, “Determination of Suitable Locations for
Power System Stabilizers and Static Var Compensators for Damping Electro-mechanical Oscillation in large Scale Power Systems,” IEEE
Trans. on Power Systems, Vol. 5, No.4, pp. 1455-1469, November
1990. [27] S. N. Singh and A. K. David, “Placement of FACTS Devices in Open
Power Market,” Proceeding of the 5th International Conference on
Advance in Power System Control, Operation and Management, APSCOM 2000, Hong Kong, pp. 173-177, October 2000.
[28] S. N. Singh and A. K. David, “A new approach for placement of
FACTS Devices in Open Power Market,” IEEE Power Engg. Rev.,
2001, 21(9), pp. 58-60. [29] M. K. Verma, and S. C. Srivastava, “Optimal Placement of SVC for
Static and Dynamic Voltage Security Enhancement,” International
Journal of Emerging Electric Power Systems, Vol.2, issue-2, 2005. [30] J.G. Singh, S. N. Singh, and S. C. Srivastava, “Placement of FACTS
Controllers for Enhancing Power System Loadability”, IEEE Trans.
on Power Delivery, Vol. 12, No.3, July 2006. [31] Fadela Benzergua, Abdelkader Chaker, Mounir Khiat, and Naima
Khalfallah, “Optimal Placement of Static VAR Compensator in Algerian Network,” Information Technology Journal, Vol.6, No.7,
pp.1095-1099, 2007.
[32] S.M. Alamelu, and R. P. Kumudhini Devi, “Novel Optimal Placement UPFC Based on Sensitivity Analysis and Evolutionary Programming
,” Journal of Engineering and Applied Sciences, Vol.3, No.1, pp.59-
63, 2008. [33] Srekanth Reddy Donapati, and M. K. Verma, “An Approach for
Optimal Placement of UPFC to enhance Voltage Stability Margin
under Contingencies,” Fifteenth National Power Systems Conference
(NPSC), IIT Bombay, December 2008 pp.441-446.
[34] R. K. Verma, “Control of Static VAR systems for Improvement of
Dynamic Stability and Damping of Torsional Oscillations,” Ph. D. Thesis, IIT Kanpur, April 1998.
[35] T. Orfano Gianni and R.Bacher, “Steady-State Optimization in Power
Systems with Series FACTS Devices,” IEEE Trans. on Power Systems, Vol.18, No.1, pp.19-26, February 2003.
[36] W. H. Litzenberger, R. K. Verma and J. D. Flanagan, “An Annotated
Bibliography of HVDC transmission and FACTS Devices 1996-1997,” Electric Power Research Institute, WO-3022 06 , June 1998.
[37] H. Mehta, T. W. Cease, L. Gyugyi, and C. D. Schauder, “Static
Condenser for FACTS,” FACTS EPRI Workshop, Boston, Massachusetts, pp.18-22, May 1992.
[38] Shu Liu, A. R., Messina, and Vijay Vittal, “Assessing Placement of
Controllers and Non-linear Behaviour Using Normal Form Analysis,” IEEE Trans on Power Systems, Vol.20, No.3, August 2005.
[39] Roberto Mingues, Federico Milano, Rafael Zarate-Minano, and Antonio
J. Conejo, “Optimal Network Placement of SVC Devices,” IEEE Trans. on Power Systems, Vol.22, No.4, November 2007.
[40] Mahbube Zaraatzade, Ivana Kockar, and Yong-Hua Song, “Minimizing
Balancing Market Congestion Re-dispatch Cost by Optimal Placement of FACTS Devices,” Power Tech., 2007, IEEE Lausanne, 1-5
July,2007 pp.873-878.
[41] Flavio G. M. Lima, Francisco D. Galiana, Ivana Kockar, and Jorge Munoz, “Phase Shifter Placement in Large Scale Systems Via Mixed
Integer Linear Programming,” IEEE Trans. on Power Systems, Vol.
18, No. 3, August 2003. [42] Naoto Yorino, E. E. El-Araby, Hiroshi Sasaki, and Shigemi Harada, “A
new Formulation for FACTS Allocation for Security Enhancement
against Voltage Collapse,” IEEE Trans. on Power Systems, Vol. 18, No.1, February 2003.
[43] Nuraddeen Magaji, and M. W. Mustafa, “Optimal Location of FACTS
devices for damping oscillations using Residue Factor,” 2nd IEEE International Conference on power and Energy (PECon 08),
December 1-3, 2008, Johor Baharu, Malaysia.
[44] E. J. Oliveira, J. W. M. Lima, and K de. Almeida, “Allocation of FACTS devices in Hydrothermal Systems,” IEEE Trans. on Power
Systems, Vol. 15, No.1, pp. 276-282, February 2000.
[45] C. S. Chang, and J. S. Huang, “Optimal SVC Placement for Voltage Stability Reinforcements,” Electric Power System Research, Vol.42,
pp.165-172, 1997.
[46] G. Cell, and F. Pilo, “Optimal Sectionalizing Switches Allocations in Distribution Networks,” IEEE Trans. on Power Delivery, Vol. 14, No.
3, July 1999. [47] Tjing T. Lie and Wanhong Deng, “Optimal Flexible AC Transmission
Systems (FACTS) Devices Allocation,” Electrical Power & Energy
Systems, Vol. 19, No.2, pp. 125-134, 1997. [48] Zuwei Yu, and D. Lusan, “Optimal Placement of FACTS Devices in
Deregulated Systems Considering line losses,” Electrical Power and
Energy Systems, Vol. 26, pp. 813-819, 2004.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-8
[49] P. Pateni, S. Vitet, M. Bena, and A.Yokoyama, “Optimal Location of
Phase Shifters in the French Network by Genetic Algorithm,” IEEE
Trans. on Power Systems, Vol. 14, No. 1, pp. 37-42, February 1999. [50] S.Gerbex, R. Cherkaoui, and A.J.Germond, “Optimal Location of Multi-
Type FACTS Devices in a Power System by Means of Genetic
Algorithms,” IEEE Trans. on Power Systems, vol.16, No.3, pp. 537-544, August 2001.
[51] K. Vijakumar, and R.P.Kumudinidevi, “A New Method for Optimal
Location of FACTS Controllers Using Genetic Algorithm,” Journal of Engineering and Applied Sciences, 2(10), pp.1576-1580, 2007.
[52] Mohd WAzir Mustafa, and Wong Yan Chiew, “Optimal Placement of Static VAR Compensator Using Genetic Algorithms,” Elektrika, Vol.
10, No.1, pp.26-31, 2008.
[53] S. N. Singh, and I. Erlich, “ Locating Unified Power Flow Controllers for Enhancing Power System Loadability,” Future Power Systems, 2005,
International Conference , 18 Nov 2005, pp 5-5.
[54] H. C. Leung, and T. S. Chung, “Optimal Placement of FACTS Controllers in Power Systems by a Genetic Based Algorithm,” IEEE
1999 International Conf. on Power Electronics and Drive Systems,
PEDS’99, July 1999, Hong Kong.
[55] Lijun Cai, Istvan Erlich, Georgios Stamtsis, and Yicheng Luo, “Optimal
Choice and Allocation of FACT Devices in Deregulated Electricity
Market Using Genetic Algorithms,” Bulk Power Systems Dynamics and Control, VI, August 22-27, 2004, Cortinad’ Ampezzo, Italy.
[56] H.I. Shaheen, G. I. Rashed, and S.J. Cheng, “Optimal Location and
Parameters Setting of UPFC based on GA and PSO for Enhancing Power System Security Under Single Contingencies,” Power and
Energy Society General Meeting-Convesion and delivery of Electrical
Energy in The 21st Century, 2008,IEEE, 20-24 July, 2008, pp.1-8. [57] L. Ippolito, “ A novel Strategy for Selection of the Optimal Number and
Location of UPFC devices in Deregulated Electric Power Systems,”
Power Tech, 2005, IEEE Russia, 27-30 June, 2005, pp.1-9. [58] K. Vijayakumar, and R. P. Kumudinidevi, “A New Method for Optimal
Location of FACTS Controllers Using Genetic Algorithms,” Journal
of Theoretical and Applied Information Technology, 2005-2007 JATIT.
[59] Hassan Barati, Mehdi Ehsan, and Mahmud Fotuhi-Firuzabad, “Location
of Unified Power Flow Controller and its Parameters Setting for Congestion Management in Pool Market Model Using Genetic
Algorithm,” Power Electronics, Drives and Energy Systems, 2006,
PEDES’06, International conference , 12-15 Dec., 2006, pp 1-7. [60] Songpakit Kaewniyompanit, Yasunori Mitani, and Kiichiro Tesuji, “A
method of micro-GA Combined Neighboring Search for Approaching
to an Optimal Type Selection and Placement of a FACTS device for Power System Stabilizing Purpose in a Multi-machine power
System,” 2004 International Conference on Power Systems
Technology-POWERCON 2004, Singapore, 21-24 November, 2004, [61] L.J. Cai, I. Erlich, and G. Stamtsis, “Optimal Choice and Allocation of
FACTS Devices in Deregulated Electricity Market Using Genetic
Algorithms,” In Proc., 2005, IEEE Power Engg. Soc., General Meeting, pp. 201-207.
[62] E. E. El-Araby, N. Yorino, and H. Saraki, “A Comprehensive Approach
of FACTS Devices Optimal Allocation to Mitigate Voltage Collapse,” In Proc., 2002, IEEE Transmission and Distribution Conf. Exhib,
pp.62-67.
[63] Z. Y. Dong, Y. Wang, D. J. Hill, and Y. V. Makarov, “ A New Approach to Power Systems VAr Planning Aimed at Voltage Stability
Enhancement with Feedback Control,” In Proc. 1999, Electric Power
Engineering Power Tech.Budapest’99, pp.33-39. [64] L. Ippolito, and P. Siano, “Selection of Optimal Number and Location
of Thyristor Controlled Phase Shifters Using Genetic Based
Algorithms,” In Proc., Inst. Elect. Engg., Gen., Transm., Distrb., Vol. 151, No.5, pp.630-637, Sept.2004.
[65] P.Bhasaputra and W. Ongsakul, “ Optimal Placement of Multi-Type FACTS Devices by Hybrid TS/SA Approach,” Circuits and Systems,
2003, International Symposium on , vol.3, 25-28 May, 2003, pp III
375-III 378. [66] H. Mori, and Y. Goto, “Parallel Tabu Search Based Method for
Determining Optimal Allocation of FACTS in Power Systems,” In
Proc.,Power Conf. on Power System Technology, 2000, pp.1077-1082.
[67] W.L.Fang, and H. W. Ngan, “Optimizing location of UPFC using the
Methods of Augmented Lagrange Multipliers,” IEE Proc.,Gener.,
Transm.,Distrib.,Vol 146, No.5, Sep. 1999. [68] J. Hao, L. B. Shi, and Ch. Chen, “Optimizing location of UPFC by
means by Improved Evolutionary Programming,” IEE Proc.,Gener.,
Transm., Distrib., Vol 151, No.6, November 2004. [69] M. Gitizadeh, and M. Kalanta, “A New Approach for Congestion
Management via Optimal Location of FACTS Devices in Deregulated
Power Systems,” DRPT 2008, 6-9 April 2008, Nanjing China. [70] S. Gerbex, R. Cherkaui, and A. J. Germond, “ Optimal Location of
FACTS Devices to Enhance Power Systems Security,” In Proc., 2003, IEEE Bologria Power Tech. Conf., pp.23-26.
[71] K. Chandrasekaran, K. Arul Jey Araj, L. Sahayasenthamil, and M.
Saravanan, “A New Method to Incorporate FACTS Devices in Optimal Power Flow Using Particle Swarm Optimization,” Journal of
Theoretical and Applied Information Technology, 2005-2009, JATIT.
[72] G. I Rashed, H. I. Shaheen, and S. T. Cheng, “Optimal Location and Parameter Setting of TCSC by Both Genetic Algorithm and Particle
Swarm Optimization,” 2007, IEEE Second conf. on Industrial
Electronics and Applications, China.
[73] Adel Moradi, and M. Fotuhi-Firuzabad, “Optimal Switch Placement in
Distribution Systems Using Trinary Particle Swarm Optimization
Algorithm,” IEEE Trans. on Power Delivery, Vol.23, No. 1, January 2008.
[74] L. Gyugi, N. G. Hingorani, P. R. Nannery, and N. Tai, “Advanced Static
VAR Compensator Using Gate Turn-off Thyristor for Utility Applications,” CIGRE,23-203, 1990 Session, Paris.
[75] T. S. Dillon, “Artificial Neural Network Applications to Power Systems
and Their Relationship to Symbolic Methods,” Electrical Power &Energy Systems., Vol. 13, No.2, pp.66-72, 1991.
[76] P. K. Das, S. Saha, and P.K. Nanda, “Artificial Neural Network
Approach for Capacitor Placement in Power Systems,” In Proc., Ist. Int. Forum on Application of Neural Networks to Power Systems,
1991, pp.247-250.
[77] Hamid Falaghi, M. Reza Halhifam, and Chanan Singh, “Ant Colony Optimization Based Method for Placement of Sectionalizing Switches
in distribution networks Using a Fuzzy Multi- Objective Approach,”
IEEE Trans. on Power Delivery, Vol. 24, No. 1, January 2009. [78] J. F. Gomez, H. M. Khodr, P. M. De Oliveira, L. Ocqul, J. M. Yusta, R.
Villasana, and A. J. Urdaneta, “Ant Colony System Algorithm for the
Planning of Primary Distribution Circuits,” IEEE Trans on Power Systems, Vol. 19, No.2, May 2004.
[79] J. C.Carlisle, and A. a. El Keib, “A Graph Search Algorithm for
Optimal Placement of Fixed and Switched Capacitors on Radial Distribution System,” IEEE Trans. on Power Delivery, Vol. 15, No. 1,
January 2000.
[80] M Saravanan, S. Mary Raja Slochanal, P. Venkatesh, P. S. Abraha, “Application of PSO Technique for Optimal Location of FACTS
Devices Considering Systems Loadability and Cost of Installation,”
Power Engineering Conference, 2005, IPEC 2005, The 7th International Conference, 29 Nov 2005- 2 Dec 2005, pp 716-721.
[81] K. Visakha, D. Thukaram, L. Jenkins, and H.P. Khincha, “Selection of
UPFC Suitable Locations for System Security Improvement Under Normal and Network Contingencies,” TENCON 2003, Conference on
Convergent Technologies for Asia –Pacific Region, Vol-2, 15-17 Oct,
2003, pp755-760. [82] D. Thukarana, L. Jenkins, and K. Visakha, “Improvement of System
Security with UPFC at Suitable Locations Under Network
Contingencies of Interconnected Systems,” IEE Proc., Gener., Transm., Distrib., Vol. 152, No.5, Sep 2005.
[83] Jing Zhang, J. Y. Wen, S. J. Cheng, and Jia Ma, “A Novel SVC
Allocation Method for Power System Voltage Stability Enhancement by Normal Forms of Diffeomorphism,” IEEE Trans. on Power
Systems, Vol. 22, No. 4, November 2007. [84] M. Santiago-Luna, and J. R. Cedeno-Maldonado, “Optimal Placement of
FACTS Controllers in Power Systems via Evolution Strategies,” 2006
IEEE PES Transmission and Distribution Conf. and Exposition Latin, America, Venezuela.
[85] Esmat Rashedi, Hossien Nezamabadi-Pour, Saeid Saryazdi, ansd Malihe
M. Farsangi, “Allocation of Static VAR Compensator Using Gravitational Search Algorithm,” First Joint Congress on Fuzzy and
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-9
Intelligent Systems Ferdowsi University of Mashhad, Iran, 29-31
August 2007.
[86] R.P.Kalyani, M. L. Crow, and D. R. Tauritz, “Optimal Placement and Control of UPFC devices Using Evolutionary Computing and
Sequential Quadratic Programming,” Power Systems Conference and
Exposition 2006, PSCE’06, 2006 IEEE PES, 29 Oct.,2006- 1 Nov.,2006, pp 959-964.
[87] Shu Liu, A.R. Messina, Vijay Vithal “A Normal Form Analysis
Approach to Siting Power System Stabilizers (PSS) and Assessing Power System Nonlinear Behavior,” IEEE Trans on Power
Systems, Vol.21, No.4, November 2006.
[88] Dheeman Chatterjee and Arindam Ghosh, “Application of Trajectory
Sensitivity for the Evaluation of the Effect of TCSC Placement on Transient Stability,” International Journal of Emerging Electric Power
Systems, Vol. 8, Issue 1, 2007.
[89] U.P.Mhaskar, and A. M. Kulkarni, “Power Oscillation Damping Using FACTS Devices: Modal Controllability, Observability in Local
signals, and Location of Transfer Function Zeros,” IEEE Trans on
Power Systems, Vol.21, No.1, February 2006.
[90] Seungwon An, John Codren and W.Gedra, “An Ideal Transformer UPFC
Model, OPF First-Order Sensitivities, and Application to Screening
for Optimal UPFC Locations”, IEEE Trans on Power Systems, Vol.22, No.1, February 2007.
[91] S.M. Alamelu, and R. P. Kumudhini Devi, “Novel Optimal Placement of
UPFC Based on Sensitivity Analysis and Evolutionary Programming,” Journal of Engineering and Applied Sciences 3(1),
59-63, 2008.
[92] V.Ajjarapu, Ping Lin Lau, and S. Battula, “An Optimal Reactive Power Planning Strategy Against Voltage Collapse,” IEEE Trans on Power
Systems, Vol.9, No.2, May 1994.
[93] T. Jain, S. N. Singh, and S.C. Srivastava, “Dynamic ATC Enhancement Through Optimal Placement of FACTS Controllers,” Electric Power
System Research, Vol. 79, Issue 11, May 2009, pp.1473-1482.
[94] Srinivasa Rao Pudi, and S.C.Srivastava, “Optimal Placement of TCSC Based on A Sensitivity Approach for Congestion Management,”
Fifteenth National Power Systems Conference (NPSC), IIT Bombay,
December 2008, pp.558-563. [95] C. A. Canizares, A. Berizzi, and P. Marannino, “Using FACTS
Controllers to Maximize Available Transfer Capability,” Bulk Power
Systems Dynamics and Control, IV-Restructuring, August 24-28, 1998, Santorim, Greece.
[96] L. Rouco, and F.L. Pagola,“An Eigen-value Sensitivity Approach to
Location and Controller Design of Controllable Series Capacitors for Damping Power System Oscillations,” IEEE Trans on Power
Systems, Vol.12, No.4, November1997.
[97] J. G. Singh, S. N. Singh, and S. C. Srivastava, “Optimal Placement of Unified Power Flow Controller Based on System Loading
Distribution Factors,” Electric Power Components and Systems, Vol.
37, Issue 4, April 2009, pp.441-463 [98] Nuraddeen Magaji, and M. W. Mustafa, “Optimal Location of TCSC
Device for Damping Oscillations,” ARPN Journal of Engineering and
Applied Sciences, Vol. 4, No.3, May 2009. [99] M. W. Mustafa, Nuraddeen Magaji, and “Optimal Location of Static Var
Compensator Device for Damping Oscillations,” American Journal of
Engineering and Applied Sciences, Vol. 2, No.2, 2009. [100] Ashwani kumar Sharma,“Optimal Number and Location of TCSC and
Loadability Enhancement in Deregulated Electricity Markets Using
MINLP,” International Journal of Emerging Electric Power Systems, Vol.5, Issue 1, 2006.
[101] Guang Ya Yang, Geir Hovland, Rajat Mujumder, and Zhao Yang
Dong, “TCSC Allocation Based on Line Flow Based Equations Via Mixed-Integer Programming,” IEEE Trans on Power Systems,
Vol.22, No.4, November 2007 [102] A hybrid Meta-Heuristic Method for Optimal Allocation of UPFCs,”
Circuits and Systems, 2009, ISCAS 2009, IEEE International
Symposium , 24-27 May, 2009, pp 1705- 1708. [103] Tina Orfanogianni, and Rainer Bacher, “Steady-State Optimization in
Power Systems with Series FACTS Devices,” IEEE Trans on Power
Systems, Vol.18, No.1, February 2003.
[104] F.F. Moghaddam, H.Nezamabadipour, and Malihe M. Farsangi,
“Curved Space Optimization for Allocation of SVC in a Large
Systems,” in Proc. Of the 6th Conference on Application of Electrical Engineering, Istanbul, Turkey, 2007, pp.59-64.
[105] Baskaran Jeevarathinam, “Genetic Algorithm and Fuzzy Logic Based
Optimal Location of FACTS Device in a Power System Network,” International Journal of Emerging Electric Power Systems, Vol. 5,
Issue 2, 2006.
[106] Lucio Ippolito, Antonio La Cortiglia, and Michele Petrocelli, “Optimal Allocation of FACTS Devices by Using Multi-Objective Optimal
Power Flow and Genetic Algorithms,” International Journal of Emerging Electric Power Systems, Vol. 7, Issue 2, 2006.
[107] Yuan-Lin Chen, “Weak Bus-Oriented Optimal Multi-Objective VAR
Planning,” IEEE Trans on Power Systems, Vol. 11, No.4, November 1996.
[108] Ying-Yi Hong, and Chen-Ching Liu, “A heuristic and Algorithmic
Approach to VAR Planning,” IEEE Trans on Power Systems, Vol. 7, No.2, May 1992.
[109] Toan Dang Nguyen, Didier Georges, and Quoc Tuan Tran, “An Energy
Approach for the Placement of Controllers & Sensors in a Power
Systems,” International Journal of Emerging Electric Power Systems,
Vol. 9, Issue 6, 2008.
[110] L. J. Cai, I. Erlich, and G. Stamtsis, “Optimal Choice and Allocation of FACTS Devices in Deregulated Electricity Market Using Genetic
Algorithms,” Power Systems Conference and Exposition, 2004, IEEE
PES,Vol-1, 10-13 OCT, 2004, pp 201-207. [111] C. P. Gupta, “Voltage Stability Margin Enhancement using FACTS
controllers,” Ph. D. Thesis, IIT Kanpur, October, 2000.
BIOGRAPHIES
Bindeshwar Singh was born in Deoria, U.P., India, in 1975. He received the B.E. degree in electrical engineering from the Deen Dayal of University of Gorakhpur, Gorakhpur, U.P., India, in 1999, and M. Tech.
in electrical engineering (Power Systems) from the Indian Institute of
Technology (IITR), Roorkee, Uttaranchal, India, in 2001. He is now a Ph. D.
student at Uttar Pradesh Technical University, Lucknow, U.P., India. In 2001, he joined the Department of Electrical Engineering, Madan Mohan Malviya Engineering College, Gorakhpur, as an Adoc. Lecturer. In 2002, he joined the Department of Electrical Engineering, Dr. Kedar Nath Modi Institute of Engineering & Technology, Modinagar, Ghaziabad, U.P., India, as a Sr. Lecturer and subsequently became an Asst. Prof. & Head in 2003. In 2007, he joined the Department of Electrical & Electronics Engineering, Krishna Engineering College, Ghaziabad, U.P., India, as an Asst. Prof. and subsequently became an Associate Professor in 2008. Presently, he is an Assistant Professor with Department of Electrical Engineering, Kamla Nehru Institute of Technology, Sultanpur, U.P., India, where he has been since August’2009.
His research interests are in Placement and Coordination of FACTS
controllers in multi-machine power systems and Power system Engg. Mobile: 09473795769, 09453503148 Email:[email protected]
Nikhlesh Kumar Sharma received the Ph.D. in electrical engineering from
the Indian Institute of Technology, Kanpur, in 2001. Currently, he is a Director with, Raj Kumar Goel Engineering College, Pilkhuwa, Ghaziabad, U.P., India,
where he has been since June’2009. His interests are in the areas of FACTS control and Power systems.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0205-10
Mobile: 09654720667, 09219532281
Email: [email protected]
A.N.Tiwari received the Ph.D. in electrical engineering from the Indian
Institute of Technology, Roorkee, in 2004. Currently, he is an Asst. Prof. with
Department of Electrical Engineering, Madan Mohan Malviya Engineering College, Gorakhpur,U.P., India, where he has been since June’1998. His
interests are in the areas of Electrical Drives and Application of Power
Electronics. Mobile: 09451215400
Email:[email protected]
1
Praveen Kumar Shukla*, Surya Prakash Tripathi**
Abstract – In real world computing environment, the information is not complete, precise and certain, making very difficult to derive an actual decision. To deal with processing and modeling of such information, fuzzy techniques are applied to exercise the proper conclusion. This paper focuses on the basics of Fuzzy Logic and its application in Rule Based Systems to make them capable to handle the real world problems. Also, different research issues associated with FRBSs have been discussed.
Keywords – Fuzzy Logic (FL), Fuzzy Sets, Linguistic Variables, Control Systems, Fuzzy Rule Based Systems (FRBS).
I. INTRODUCTION Humans are capable to use linguistic information
precisely in their decision making. Due to imprecise and uncertain nature of the linguistic information, machines are not capable to use them in decision making processes using traditional methods. To make the machines intelligent, like humans in this regard, Fuzzy Techniques are used.
The idea of the Fuzzy Logic was first introduced by Professor Lotfi Ahmad Zadeh, at University of Berkeley, California in his seminal paper “Fuzzy Sets” [1].
Fuzzy Logic [2, 3] is a form of multi-valued logic derived from fuzzy set theory to deal with approximate reasoning. It provides the means to represent and process the linguistic information and subjective attributes of the real world. Fuzzy Logic is the extension of Boolean Crisp Logic to deal with the concept of partial truth. Fuzzy Logic is applied in the number of areas, i.e. engineering applications, medical applications, economics and management, industrial applications and many more. It is also integrated with other soft computing techniques, like Neural Network (an approach that mimics the functionality of human brain), Genetic Algorithms (a nature inspired search and optimization technique), PSO (Particle Swarm Optimization) etc. In the early stage of the Fuzzy Logic, a number of misconceptions have been created. Here we are going to introduce few points about fuzzy logic to make the concept very clear. 1. Fuzzy logic is not fuzzy. 2. Fuzzy logic is precise. 3. Fuzzy Logic is a precise system of reasoning, deduction and computation in which the objects of
discourse and analysis are associated with information ,which is or is allowed to be imperfect. 4. Any formal system can be fuzzified. Rule Base Systems [4] are highly applicable in decision making, control systems and forecasting. To deal with imprecise, uncertain and inexact real world knowledge, in rule based systems, fuzzy techniques are used. Fuzzy logic is the way to represent the complex situations in terms of simple natural languages. This paper introduces the Fuzzy Rule Based Systems (FRBS) and different research issues associated with them. In section II, the basic mathematical concept of the Fuzzy Logic has been introduced. Section III introduces the two basic types of FRBS, Mamdani FRBS and TSK FRBS. In section IV, authors revisited five research issues with the Fuzzy Rule Based Systems (FRBS). In section V, conclusion and future scope of the research issues ha been discussed.
II. BASIC CONCEPTS: FUZZY LOGIC
The theory of Fuzzy Logic [5] can be developed using the concepts of Fuzzy Sets similar to how theory of crisp bivalent logic can be developed using the concept of crisp sets. Specifically, there exists an isomorphism between sets and logic. In view of this, a good foundation of the fuzzy sets is necessary to develop the theory of Fuzzy Logic.
A fuzzy set is a set without clear or sharp (crisp) boundaries. Partial membership degree is possible in fuzzy sets. In other words, softness is associated with the membership of elements associated.
Examples may include, like TEMPERATURE. This fuzzy variable may take fuzzy values, like COLD, COOL, WARM, HOT. A fuzzy value such as ‘WARM’ is a fuzzy descriptor.
A. Universe of Discourse
If universe of discourse is represented by X, is a set that contains every set of interest in the context of a given class of a problem. The vein diagram of the fuzzy set is given in Fig. 1.
Fig. 1 Venn diagram of Fuzzy Set
Fuzzy Logic and Rule Based Systems: Research Issues & Challenges
* Senior Lecturer, Department of Information Technology, Northern India Engineering College, Lucknow ([email protected])
**Associate Professor, Department of Computer Science &
Engineering, Institute of Engineering & Technology (A Constituent College of GBTU), Lucknow ([email protected])
2
B. Representation of Fuzzy Sets There are two representation techniques of fuzzy sets,
membership function method and symbolic representation. Membership Function Method
This function gives the grade (degree) of membership within the set of any element of Universe of Discourse. The membership function maps the elements of the universe on to the numerical values in the interval [0, 1].
]1,0[:)( XxA Here, )(xA is the membership function of the fuzzy
set A in the universe X. It is defined as follows: ]1,0[)(,));(,( xAXxxAxA
The membership function represents the grade of possibility that an element x belongs to the set A. It is a possibility function, not a probability function.
Symbolic Representation
A fuzzy set may be symbolically represented as follows: )(| xAxA . They can also be represented as a formal series, when the universe is discrete in the nature.
....................../)1(
...............2/)2(1/)1(
ixixA
xxAxxAA
or
Xix ix
ixAA)(
If the universe is continuous then it can be represented as follows:
Xx x
xAA)(
C. Algebraic Operations on Fuzzy Sets
Let X is a set of objects with elements denoted by x, i.e. X=x
A fuzzy set A in X is characterized by a membership function )(xA , which maps each point in X on to the real interval [0.0, 1.0]. As )(xA approaches 1.0, the grade of membership of X in A increases.
1. A is empty iff for all x, )(xA =0.0 2. If A and B are the two fuzzy sets, then A=B iff for
all x: )(xA = )(xb .
3. )'(xA =1- )(xA
4. A is contained in B iff )(xA <= )(xb 5. AUNIONBBAC , where
))(),(max()( xbxaxC 6. IONBAINTERSECTBAC , where
))(),(min()( xbxaxC
D. Support Set This is a crisp set of a fuzzy set containing all the
elements (in the universe) whose membership grade is greater than 0. The support set S of a fuzzy set A with membership function )(xA is given by
0)(| xAXxS
Fig. 2 Fuzzy Membership Functions E. -cut of the fuzzy set
The -cut of the fuzzy set A is the crisp set denoted by A formed by those elements of A whose membership function grade is greater than or equal to a specified threshold value .
]1,0[,)(| xAXxA The strong -cut is defined by
]1,0[,)(| xAXxA When =0 then -cut will become the support set of
a fuzzy set. Also, the fuzzy AND, OR and NOT operations can be
generalized. The generalized FUZZY AND operation is called Triangular Norm (T-Norm) and generalized FUZZY UNION is called T-Conorm (S- Norm). The basic concepts of Fuzzy Set Theory may be studied in [3, 5].
III. FUZZY RULE BASED SYSTENS (FRBS)
Fuzzy Rule Based Systems (FRBS) constitute an
extension to classical rule based systems, because they deal with IF – THEN rules whose antecedents and consequents are composed of fuzzy logic statements, in place classical logical ones.
The most common applications of FRBS includes, Fuzzy Modeling [6], Fuzzy Control [7] and Fuzzy Classification [8]. In a FRBS, fuzzy logic used to perform the operations like, representation of different form of knowledge, modeling the interactions and relationships that exist among its variables. The main features of the knowledge captured by fuzzy sets involve handling of uncertainty. Due to this, inference methods have become more robust and flexible with the approximate reasoning methods using Fuzzy Logic.
Linguistic variables and values are used for the enhancement of the Knowledge Representation. These variables and their values are defined by the context dependent fuzzy sets whose meanings are specified by gradual membership function.
Two major types of FRBSs proposed are, Mamdani Fuzzy Rule Based Systems [9] and Takagi-Sugeno-Kang FRBSs [10]. A. Mamdani Fuzzy Rule – Based Systems
These are the FRBS with fuzzifier and defuzzifier, more commonly these are known as Fuzzy Logic
3
Controllers (FLCc). The major constituents of the Mamdani FRBS are shown in Figure 3.
Fig. 3 Mamdani Fuzzy Rule Based Systems
1. Knowledge Base: The Knowledge Base (KB) stores the available knowledge about the problem in the form of fuzzy “IF THEN” rules. It composed of two main components, Data Base (DB) and Rule Base (RB). Data Base (DB) stores the membership functions of fuzzy sets and scaling functions for context adaptation purpose. Rule Base (RB) stores the FUZZY IF THEN rules for the purpose inference and decision making. Multiple rules can be fired simultaneously for the same input.
2. Fuzzification Interface: It transforms the crisp input data into fuzzy values that acts as input to fuzzy reasoning process.
3. Inference System: It infers from the fuzzy input to several resulting output fuzzy sets according to the information stored in the Knowledge Base (KB).
4. Defuzzification Interface: It converts the fuzzy sets obtained from the inference process into a crisp action that constitutes the global output of the FRBS.
B. TSK Fuzzy Rule Based Systems
A new FRBS model is proposed, based on rules whose antecedent is composed of the linguistic variables and the consequent is represented by a function of the input variables.
IF X1 is A1 and…………and Xn is An THEN Y=p1.X1+……+pn.Xn+p0
Here, Xi is the system input variable, Y as the output variable p= (p0, p1,…….., pn) is a vector of real parameters. Ai is the direct specification of a fuzzy set or linguistic label that points to a one particular member of a fuzzy partition of a linguistic variable.
The output of a TSK FRBS using a KB composed of m rules is obtained as a weighted sum of the individual outputs provided by each rule, Yi is given by
mi ih
iYmi ih
1
.1
Here, i=1, 2,………,m. ))(.,),........1(1( nxinAxiATih is the matching degree
between the antecedent part of the ith rule and the current
inputs to the systems, ),.......1(0 nxxx . T stands for the conjunctive operator modeled by a t-norm.
Fig. 4 TSK Type Fuzzy Rule Based Systems
IV. RESEARCH ISSUES AND CHALLENGES
Application of fuzzy computing in rule based systems
provides a mathematical platform to deal with information imprecision in decision making processes. Several issues have been raised by the researchers in this field. We are presenting here the burning issues in the FRBS research. A. Context Adaptation
Human decisions and perceptions are extremely dependent on the context. Therefore, when Fuzzy Logic models the human decision making capabilities in machines, it is mandatory to apply context in Fuzzy Systems.
Context adaptation in Fuzzy Systems has been approached as scaling of fuzzy sets from one universe of discourse to another by means of non-linear scaling functions, whose parameters are identified by the data.
The idea behind this research issue is to develop new context aware approaches to automatic development of the fuzzy systems from data. Other objectives includes, interpretability oriented adaptation, identification of context variable, high level of rule identification.
The abstract idea of this research issue [11] is given in Fig. 4.
Fig. 5 Context Adaptation Process In [12], a method of achieving context adaptation by adjusting an initial normalized fuzzy rule based systems through the use of operators that appropriately change the representation of linguistic variables. A multi-objective evolutionary algorithm based Pareto-optimum context adapted Mamdani Type FRBSs with different trade – offs between accuracy and interpretability is investigated in [13]. In this proposal, a
4
novel index is proposed based on fuzzy ordering relations, in-order to provide a measure of interpretability. Other approaches related to context adaptation can be studied in [14, 15, 16, 17, 18]. B. Interpretability Accuracy Trade-Off
The trade – off between interpretability and accuracy is an important research issue. The definition of the accuracy is straight forward in different applications, but the definition of the interpretability is rather problematic [19].
The main purpose is to build fuzzy systems with a user controllable trade off between accuracy and interpretability. Interpretability is maintained by the structural choices regarding the type of membership functions, rules and inference mechanism. Interpretability can be maintained or enhanced during the fuzzy systems generation or obtained by post processing of the resulting data driven fuzzy systems [20, 21].
A new linguistic rule representation model [22,23] was proposed to perform a genetic lateral tuning of membership functions, based on linguistic 2- tuple representation model that allow the lateral displacement of a label considering a unique parameter. It provides reduction of search space that eases the derivation of optimal path.
A user-controllable interpretability-accuracy trade off for fuzzy systems has been discussed in [24].
The interpretability- accuracy trade off issue is discussed with multi-objective fuzzy genetics-based machine learning in [25].
C. Fuzzy Rule Selection
In high dimensional data problems [26, 27], the number of rules in the Rule Base grows exponentially as more inputs are added. Hence, it is required to have a fuzzy rule generation method. It is likely to derive fuzzy rule sets, including following types of rules; Redundant rules: The actions of these rules are covered by the other rules. Wrong rules: These are ill defined rules and perturb the systems performance. Conflicting rules: These rules worsen the system performance, when co-existing with other rules in the RB.
Other solutions for the problem of data dimensionality and rule overflow, are rule reduction methods. Two approaches are proposed for this rule reduction:
Approach 1: To combine the membership functions of two or more rules, reducing to a single one’s (Scatter partition FRBS).
Approach 2: To select the fuzzy rules, we get the rule subsets with a good cooperation from the initial RB (descriptive and scatter FRBS).
Several methods can be obtained from selection, with different search algorithms that are providing most successful combination of fuzzy rules [28,29,30].
A genetic rule selection process in order to obtain a compact and accurate fuzzy rule based classification systems is discussed in [31].
D. Optimization of membership function and scaling function
Optimization of the membership function results in the improvement of interpretability and accuracy of the fuzzy systems. Also, the scaling functions are optimized to maintain and precise the context related issues in the fuzzy rule based systems. Scaling functions apply the universe of discourse of the input and output variables to the domain where fuzzy sets are defined. Their tuning and adaptation allows the scaled universe of discourse to match the variable range in a better way. Several parameters are considered for the purpose of optimization. These may include, 1. Scaling Functions, 2. Upper and lower bounds (Linear Scaling Functions), 3. Contraction/dilation operators (Nonlinear scaling functions).
Also the optimization of the membership functions is an important research issue. The tuning process [32] slightly adjusts the shapes of the membership functions of the preliminary data base definition. Different types of membership functions are considered for this purpose, i.e. 1. Triangular, 2. Trapezoidal, etc.
For the purpose of semantic interpretability of linguistic fuzzy models, an index is proposed [33]. Tuning of the membership function and rule selection is performed using a multi-objective evolutionary algorithms.
E. Fuzzy Partition Granularity
Obtaining good uniform fuzzy partition granularity [32] that improves the FRBS behavior is an important research issue. The granularity selection plays an important role in many characteristics of the FRBS, such as the accuracy in fuzzy modeling or the smoothening in fuzzy control. Also, the granularity of the input variables specifies the maximum number of fuzzy rules that may compose the Rule Base (RB). So, it has a strong influence on aspects, like complexity of rule learning, interpretability of the FRBS obtained or its accuracy.
The issue of the fuzzy partition granularity is a fuzzy rule base classification systems are discussed in [35].
V. CONCLUSION & FUTURE SCOPE
Modern engineering, medical and business applications
are requiring to enhance their capability to deal with imprecise and uncertain information, enabling them to have a strong reasoning and decision power. It makes them to handle more complex and linguistic computations easily and efficiently. All these requirements lead to rapid development and integration of Fuzzy Logic in control systems.
In future, authors would like to implement the solutions for the problems addressed in the section IV, by Evolutionary Computation techniques. The use of multi-objective Evolutionary Algorithms and Memetic Genetic Algorithms will be preferred.
REFERENCES
[1] L. A. Zadeh, “Fuzzy Sets”, Information and Control, Vol. 8, pp.
338-353, 1965. [2] L. A. Zadeh, “Fuzzy sets as a basis of possibility”, Fuzzy Sets
Systems, Vol. 1, pp. 3-28, 1978. [3] T. J. Ross, “Fuzzy Logic with Engineering Applications”,
McGraw-Hill, 1995.
5
[4] L. M. Pant, A. Ganju, “Fuzzy Rule Based Systems for prediction of direct action avalanche”, Current Science, Vol. 87, No.1, July, 2004.
[5] F. O. Karray, C. De Silva, “Soft Computing and Intelligent Systems Design-Theory, Tools and Applications”, Pearson Publications, 2004.
[6] W. Pedrycz (Eds.), Fuzzy Modelling: Paradigms and Practice, Kluwer Academic Press, 1996.
[7] D. Drainkov, H. Hellendorn, M. Reinfrank, An introduction to Fuzzy Control, Springer-Verlag, 1993.
[8] Z. Chi, H. Yan, T. Pham, Fuzzy Algorithms: With applications to image processing and pattern recognition, World – Scientific, 1996.
[9] E. H. Mamdani, S. Assilian, “An experiment in linguistic synthesis with fuzzy logic controllers”, International Journal of Man-Machine Studies, Vol. 7, pp. 1-13, 1975,.
[10] T. Takagi, M. Sugeno, “Fuzzy identification of systems and its applications to modeling and control”, IEEE Transactions Systems, Man and Cybernetics, Vol. 15, No. 1, pp. 116-132.
[11] URL: “http://cig.iet.unipi.it/cig/research02.html” [12] A. Botta, B. Lazzerini, F. Marcelloni, “Context adaptation in
Mamdani Fuzzy Systems through new operators tuned by a genetic algorithms”, FUZZ-IEEE, pp. 1641-1648, 2006.
[13] A. Botta, B. Lazzerini, F. Marcelloni, D. C. Stefanescu, “Context adaptation of fuzzy systems through a multi objective evolutionary approach based on a novel interpretability index”, Soft Computing, 2008, pp. 437-449.
[14] A. Botta, B. Lazzerini, F. Marcelloni, “Context adaptation of Mamdani fuzzy rule based systems”, International Journal of Intelligent Systems, Vol. 23, No. 4, pp. 397-418, 2008
[15] W. Pedrycz, R. R. Gudwin, F. A. C. Gomide, “Non linear context adaptation in the calibration of fuzzy sets”, Fuzzy Sets and Systems, Vol. 88, No 1, pp. 91-97, 1997.
[16] L. Magdalena, “On the role of context in hierarchal fuzzy controllers”, International Journal of Intelligent Systems, Vol. 17, No. 5, pp. 471-493, 2002.
[17] R. R. Gudwin, F. A. C. Gomide, W. Pedrycz, “Context adaptation in fuzzy processing and genetic algorithms”, International Journal of Intelligent Systems, Vol. 13, No. 10-11, pp. 929-948, 1998.
[18]O. Cordon, F. Herrera, L. Magdalena, P. Villar, “A genetic learning process for the scaling factors, granularity and contexts of the fuzzy rule based systems data base”, Information Science, Vol. 136 (1-4), pp. 85-107, 2001.
[19] R. Mikut, J. Jakel, L. Grall, “Interpretability issues in data based learning of the fuzzy systems”, Fuzzy Sets and Systems, Vol. 150, pp. 179-197, 2005.
[20] U. Bodenhofer, P. Bauer, A formal model of interpretability of linguistic variables, in: J. Cassilas, O. Cordon, F. Herrera, L. Magdalena (Eds.), Trade off between accuracy and interpretability in fuzzy rule based modeling, Studies in Fuzziness and Soft Computing, Physica, Heidelberg, 2002.
[21] O. Cordon, F. Herrera, “A proposal for improving the accuracy of the linguistic modeling”, IEEE Transactions on Fuzzy Systems, Vol. 8, No. 3, pp. 335-344, 2000.
[22] R. Alcala, J-Alcala-Fdez, F. Herrera, J. Otero, “Genetic learning of knowledge bases of a fuzzy system by using the linguistic 2-tuple representation”, FUZZ-IEEE, pp. 797-802, 2005.
[23] R. Alcala, J-Alcala-Fdez, F. Herrera, J. Otero, “Genetic learning of accurate and compact fuzzy rule based systems based on the 2-tuple representation”, International Journal of Approximate Reasoning, Vol. 44, pp. 45-64, 2007.
[24] R. Mikut, J. Jakel, L. Groll, “Interpretability issues in data-based learning of fuzzy systems”, Fuzzy Sets and Systems, Vol. 150, pp. 179-197, 2005.
[25] H. Ishibuchi, Y. Nojima, “Analysis of interpretability-accuracy trade off of fuzzy systems by multi-objective fuzzy genetics-based machine learning”, International Journal of Approximate Reasoning, Vol. 44, pp. 4-31, 2007.
[26] H. Ishibuchi, K. Nozaki, N. Yamamoto, H. Tanaka, “Selecting fuzzy if-then rules for classification problems using genetic algorithms”, IEEE Transaction on Fuzzy Systems, Vol. 3, No. 3, pp. 260–270, 1995.
[27] O. Cordón, F. Herrera, “A three-stage evolutionary process for learning descriptive and approximate fuzzy logic controller knowledge bases from examples”, International Journal of Approximate Reasoning, Vol. 17, No. 4, pp. 369–407, 1997.
[28] O. Cordon, F. Herrera, “A proposal for improving the accuracy of linguistic modeling”, IEEE Transactions on Fuzzy Systems, Vol. 8, No. 3, pp. 335-344, 2000.
[29] H. Ishibuchi, T. Murata, I. B. Tarksen, “Single objective and two objective genetic algorithms for selecting fuzzy rules for pattern classification problems”, Fuzzy Sets and Systems, Vol. 89, No. 2, pp. 135-150, 1997.
[30] H. Isibuchi, K. Nozaki, N. Yamamoto, H. Tanaka, “Selecting fuzzy if then rules for classification problems using genetic algorithms”, IEEE Transactions on Fuzzy Systems, Vol. 3, No. 3, pp. 260-270, 1995,.
[31] A. Fernandez, M. J. del Jesus, F. Herrera, “Analyzing the hierarchal fuzzy rule based classification systems with genetic rule selection, International workshop on genetic and evolutionary fuzzy systems, Spain, pp. 69-74, March, 2010,.
[32] O. Cordon, F. Herrera, F. Hoffmann, L. Magdalena, Genetic Fuzzy Systems: Evolutionary Tuning and Learning of Fuzzy Knowledge Bases, Applications in Fuzzy Systems – Applications and Theory, World Scientific, Vol. 19, 2001.
[33] M. J. Gacto, R. Alcala, F. Herrera, “Integration of an index to preserve the semantic interpretability with multi-objective evolutionary rule selection and tuning of linguistic fuzzy systems”, IEEE Transactions on Fuzzy Systems, Vol. 18, No, 3, , pp. 515-531, June 2010.
[34] O. Cordon, F. Herrera, P. Villar, “Analysis and guidelines to obtain a good uniform fuzzy partition granularity for fuzzy rule based systems using simulated annealing”, International Journal of Approximate Reasoning, Vol. 25, pp. 187-215, 2000.
[35] A. Fernandez, S. Garcia, M. J. del Jesus, F. Herrera, “A study of the behavior of linguistic fuzzy rule based classification system in the framework of imbalanced data sets”, Fuzzy Sets and Systems, Vol. 159, pp. 2378-2398, 2008.
Praveen Kumar Shukla is with the department of Information Technology, Northern India Engineering College, Lucknow. He is working here as a Senior Lecturer and is the research scholar in the Department of Computer Science & Engineering, Gautam Buddh Technology University (Formerly UPTU), Lucknow. He is B. Tech. in Information Technology and M. Tech. in Computer Science. He is working on Genetic Fuzzy Systems. His area of interest includes, Computational Intelligence
techniques (Fuzzy Logic, Neural Network, Genetic Algorithms) and other Nature Inspired Artificial Intelligence Techniques. He has published many papers in different National/International Conferences and Journals.
Dr. Surya Prakash Tripathi is with the
Department of Computer Science & Engineering, Institute of Engineering & Technology (A constituent college of GBTU), Lucknow as an Associate Professor. He is M. Tech. from IIT, Delhi and Ph. D. from Lucknow University, Lucknow. He has guided many Ph.D. scholars. His area of interest includes, Databases Operating Systems, Soft Computing Techniques and other Nature Inspired Artificial Intelligence
techniques. He has published many papers in National/International Conferences/ Journals.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-1
A comparative analysis of controllers controlling uncertain factors affecting the
robust position control of DC motor
Farhad Aslam Gagandeep Kaur
Deptt. of Electrical & Instrumentation Engg. Deptt. of Electrical and Instrumentation Engg.
Thapar University, Patiala. Thapar University, Patiala.
[email protected] [email protected]
ABSTRACT
All the industrial process applications require
robust position control of DC motor . The aim
of this paper is to design a robust position
control of DC motor by selecting different
controllers like P, PI, PID and their tuning
methods. The model of a DC motor is
considered as a third order system with
incorporating uncertainty. This paper
compares the different kinds of tuning
methods of parameter for PID controller. One
is the controller design by Zeigler and Nichols
method, second is the auto tuning of the
controller in basic design mode and third is in
the extended design mode. It was found that
the proposed PID parameters adjustment in
the basic and extended design mode is far
better than the P, PI and Zeigler and Nichols
method. The proposed method could be
applied to the higher order system also.
Keywords: DC motor, PID tuning, Basic
mode, Extended mode, Robust position
control.
I. INTRODUCTION
Due to its excellent speed control
characteristics , the DC motor has been widely
used in industry even though its maintenance
costs are higher than the other motors like
induction motor, synchronous motor,
brushless dc motor. As a result, robust
position control of DC motor has attracted
considerable research and several method
have evolved. Proportional- Integral-
Derivative (PID) controllers along with their
tuning have been widely used for speed and
position control of DC motor.
This paper endeavors to design a system using
two methods of auto tuning of PID parameters
called Basic design mode and Extended
design mode. Auto tuning is basically used to
tune PID gains automatically in a Simulink
model containing a PID controller block. The
PID tuner allows to achieve a good balance
between performance and robustness. It
automatically computes a linear model of the
plant. The PID tuner considers the plant to be
the combination of all blocks between the PID
controller input and output. Thus, the plant
includes all blocks in the control loop, other
than the controller itself. The main objectives
of PID tuner are closed- loop stability (in
which system output remains bounded for
bounded input), adequate performance (in
which closed- loop system tracks reference
changes and suppresses disturbance as rapidly
as possible) and adequate robustness (the loop
design has enough gain margin and phase
margin to allow for modeling errors or
variations in system dynamics.
Basic design mode of PID tuner refine the
controller design by adjusting response time.
It make the closed- loop response of the
controlled system faster or slower. Extended
design mode of PID tuner refine the controller
design by separately adjust loop bandwidth
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-2
and phase margin. The larger the loop
bandwidth, the faster the controller responds
to changes in the reference or disturbances in
the loop. The larger the phase margin, the
more robust the controller is against modeling
errors or variations in plant dynamics. The
objective of this paper is to show that by
employing the proposed tuning of PID
controllers, an optimization can be achieved.
This can be seen by comparing the result of
the PID tuner against the classically tuned
system.
II. MODELING A DC MOTOR
To be modeling a DC Motor, simple circuit of
its electrical diagram as shown in Fig. 1 is
considered. To be Modeling and Simulate the
DC motor, the following steps are to be made
step by step;
Step1: Represent the DC motor
circuit diagram.
Step2: Represent system equations
Step3: Calculate the Transfer
function
Step4: Convert to model block
Step5: Run the Simulation
Step6: Analysis
A. Closed-Loop System Consideration
To perform the simulation of the system, an
appropriate model needs to be established.
Therefore, a model based on the motor
specifications needs to be obtained. Fig. 1
shows the DC motor circuit with Torque and
Rotor Angle consideration.
Fig. 1 Schematic diagram of a DC motor
B. System Equation
The motor torque T is related to the armature
current, i , by a torque constant K;
T = K i (1)
The generated voltage, ea, is relative to
angular velocity by;
ea = K wm = K dθ/dt (2)
From Fig. 1 we can write the following
equations based on the Newton‟s law
combined with the Kirchhoff‟s law:
J d2θ/dt
2 + b dθ/dt = K i (3)
L di/dt + Ri = V- K dθ/dt (4)
C. Transfer Function
Using the Laplace transform, equations (3)
and (4) can be written as:
J s2 θ(s) + b s θ(s) = K I(s) (5)
L s I(s) + R I(s) = V(s) – K s θ(s) (6)
Where s denotes the Laplace operator. From
(6) we can express I(s):
and substitute it in (5) to obtain:
This equation for the DC motor is shown in
the block diagram in Fig. 2. From equation
(8), the transfer function from the input
voltage, V (s), to the output angle, θ, directly
follows:
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-3
Before any consideration of the above
equations, we must know the constant values
of data, K, J, b, V, L and R. This is very
important to the application of DC motor
which we will be used.
Speed N = 1220 rpm = 127.7 rad/sec
Motor inertia J = 1 Kg.m2 V = 240V
EO = V – i Ra = 230.3 V Power P = EO Ia =
3731 W = 5 HP
If = V/ Rf = 1 where Rf = 240 Ω EO = w L
If = 127.7 rad/sec
Therefore for the max speed rpm of 1220, it
can be calculate the torque constant K;
K = 1.88 N. m/A
By using equation (3), for w = dθ/dt
At the steady state (used as analyzed data),
both I and w are stabilized;
Therefore, the total equivalent damping b can
be chosen the value of;
By calculating and assuming the require data
as above, the following value are assigned to
be used for our desire DC Motor Model.
V=240 V; J= 1 kg.m2; b =0.24 N.m.s;
K =1.88 N. m/A; R =0.6 Ω; and L =0.5 H;
Fig. 2 A closed- loop system that
Representing the DC motor
III. SIMULATION, TESTING AND
RESULTS
There are four cases in which uncertainty in
the form of disturbance and load is
incorporated to the system.
Our Design requirement
Maximum Overshoot = < 2%
Undershoot = 0
Rise time ( tr) = < 7 sec
Settling time ( ts) = < 12 sec
Case1: No Load and No Disturbance
Fig. 3 Process model with P controller
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-4
Fig. 4 System output with P controller
Fig. 5 Process model with PI controller
Fig. 6 System output with PI controller
Fig. 7 Process model with PID controller
Fig. 8 System output with PID controller
Fig. 9 System output with PID controller
(Zeigler and Nichols tuning method)
Fig. 10 System output with PID controller
auto tuning (Basic design mode)
Fig. 11 System output with PID controller
auto tuning (Extended design mode)
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-5
Fig. 12 Step disturbance rejection plot with
PID controller auto tuning (Extended design
mode)
Fig. 13 Open- loop Bode plot with PID
controller auto tuning (Extended design mode)
Controller
types
KP KI KD Rise
time
(tr
in
sec)
Settling
time (ts
in sec)
Over
shoot
(%)
Under
shoot
(%)
Gain
margin(db@
rad/sec)
Phase
margin
(deg@rad/sec)
P 0.5 0 0 6.19 15.6 0.377 30 [email protected] [email protected]
PI 0.5154 0.07252 0 2.68 23.9 25.2 8 [email protected] [email protected]
PID 0.1 0.00001 1.001 70 130 0 0 - -
PID (ZN) 0.0138 0.00123 0.038795 50 1500 65 33 - -
PID(Basic) 0.73195 0.0029102 -1.5315 3.36 14.2 9.85 0 [email protected] [email protected]
PID(Ext.) 0.5247 0.0023067 -1.0011 5.59 8.45 1.62 0 [email protected] [email protected]
Table 1. Comparative analysis of different controllers under no load & no disturbance
Hence after implementation of auto tuning of
PID controller in extended design mode, the
system design requirement is achieved which
is shown in table 1.
Case2: Load and No Disturbance
In this case uncertainty in the form of 2nd
order load is incorporated.
Fig. 14 Process model with P controller
Fig. 15 System output with P controller
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-6
Fig. 16 Process model with PI controller
Fig. 17 System output with PI controller
Fig. 18 Process model with PID controller
Fig. 19 System output with PID controller
auto tuning (Basic design mode)
Fig. 20 System output with PID controller
auto tuning (Extended design mode)
Fig. 21 Step disturbance rejection plot with
PID controller auto tuning (Extended design
mode)
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-7
Fig. 22 Open- loop Bode plot with PID
controller auto tuning (Extended design mode)
Controller
types
KP KI KD Rise
time
(tr
in
sec)
Settling
time (ts
in sec)
Over
shoot
(%)
Under
shoot
(%)
Gain
margin(db@
rad/sec)
Phase
margin
(deg@rad/sec)
P 0.43695 0 0 9.6 16.1 0.00758 32 [email protected] [email protected]
PI 0.35952 0.037665 0 5.45 34 23.7 0 [email protected] [email protected]
PID(Basic) 0.76098 0.0031897 -1.276 2.65 11 8.05 0 [email protected] [email protected]
PID(Ext.) 0.43964 0.0019183 -1.103 6.53 11.6 1.12 0 [email protected] [email protected]
Table 2. Comparative analysis of different controllers under load & no disturbance
After incorporating 2nd
order load, the design
requirement of the system is achieved in the
extended design mode which is shown in table
2.
Case3: No Load and Disturbance
In this case uncertainty in the form of step
disturbance of final value 0.5 is incorporated.
Fig. 23 Process model with P controller
Fig. 24 System output with P controller
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-8
Fig. 25 Process model with PI controller
Fig. 26 System output with PI controller
Fig. 27 Process model with PID controller
Fig. 28 System output with PID controller
auto tuning (Basic design mode)
Fig. 29 System output with PID controller
auto tuning (Extended design mode)
Fig. 30 Step disturbance rejection plot with
PID controller auto tuning (Extended design
mode)
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-9
Fig. 31 Open- loop Bode plot with PID
controller auto tuning (Extended design mode)
Controller
types
KP KI KD Rise
time
(tr
in
sec)
Settling
time (ts
in sec)
Over
shoot
(%)
Under
shoot
(%)
Gain
margin(db@
rad/sec)
Phase
margin
(deg@rad/sec)
P 0.41817 0 0 9.75 19.1 0 20 [email protected] [email protected]
PI 0.56269 0.084676 0 2.2 22.8 27.2 1 [email protected] [email protected]
PID(Basic) 0.43964 0.0019183 -1.1032 6.53 11.6 1.12 0 [email protected] [email protected]
PID(Ext.) 0.86883 0.0035009 -1.4425 2.37 10.9 9.9 0 [email protected] [email protected]
Table 3. Comparative analysis of different controllers under no load & disturbance
Hence after incorporating disturbance, the
design requirement fulfill in Basic design
mode which is shown in table 3.
Case4: Load and Disturbance
In this case uncertainty in the form of step
disturbance of final value 0.5 and a 2nd
order
load is incorporated.
Fig. 32 Process model with P controller
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-10
Fig. 33 System output with P controller
Fig. 34 Process model with PI controller
Fig. 35 System output with PI controller
Fig. 36 Process model with PID controller
Fig. 37 System output with PID controller
auto tuning (Basic design mode)
Fig. 38 System output with PID controller
auto tuning (Extended design mode)
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-11
Fig. 39 Step disturbance rejection plot with
PID controller auto tuning (Extended design
mode)
Fig. 40 Open- loop Bode plot with PID
controller auto tuning (Extended design mode)
Controller
types
KP KI KD Rise
time
(tr
in
sec)
Settling
time (ts
in sec)
Over
shoot
(%)
Under
shoot
(%)
Gain
margin(db@
rad/sec)
Phase
margin
(deg@rad/sec)
P 0.36613 0 0 10.4 20.3 0 20 [email protected] [email protected]
PI 0.58763 0.091329 0 2.07 22.5 28.7 5 [email protected] [email protected]
PID(Basic) 0.79159 0.0033399 -1.2026 2.47 10.7 24 3 [email protected] [email protected]
PID(Ext.) 0.4814 0.0021039 -1.1276 5.9 9.46 1.47 3 [email protected] [email protected]
Table 4. Comparative analysis of different controllers under load & disturbance
Hence after having uncertainty in the form of
both 2nd
order load and step disturbance, the
best controller performance is achieved in
extended design mode which is shown in table
4.
IV. CONCLUSION
Electric machines are used to generate
electrical power in power plants and provide
mechanical work in industries. The DC
machine is considered to be basic electric
machines. The aim of this paper is to
introduce Technicians to the modeling of
power components and to use computer
simulation as a tool for conducting transient
and control studies. Next to having an actual
system to experiment on, simulation is often
chosen by engineers to study transient and
control performance or to test conceptual
designs.
MATLAB/SIMULINK is used because of the
short learning curve that most students require
to start using it, its wide distribution, and its
general-purpose nature. This will demonstrate
the advantages of using MATLAB for
analyzing power system steady state behavior
and its capabilities for simulating transients in
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0301-12
power systems and power electronics,
including control system dynamic behavior.
This paper has demonstrated the
implementation of auto tuning of PID
controller in MATLAB/SIMULINK, both in
basic and extended design mode. This is easy
to implement and requires a small amount of
time. The controller showed robust
performance of the system under different
cases and optimization of our system is
achieved which is less complex than other
controllers and optimization technique.
V. REFERENCES
[1] Steven T.Karris, „Introduction to Simulink
with Engineering Applications‟, Orchard
Publications,www.orchardpublications.com
[2] Tan Kiong Howe, May 2003, Thesis, B.E
(Hons.), „Evaluation of the transient
response of a DC motor using
MATLAB/SIMULINK‟, University of
Queensland.
[3] Math Works, 2001, Introduction to
MATLAB, the Math Works, Inc.
[4] O. Dwyer, .PI And PID Controller Tuning
Rules for Time Delay Process: A
Summary. Part 1: PI Controller Tuning
Rules, Proceedings of Irish Signals and
Systems Conference, June1999
[5] Raghavan S. Digital control for speed and
position of a DC motor. MS Thesis, Texas
A&M University, Kingsville, 2005.
[6] M. Chow and A. Menozzi, “on the
comparison of emerging and conventional
techniques for DC motor control,” Proc.
IECON, pp. 1008- 1013, 1992.
[7] SimPowerSystems for use with Simulink,
users guide, Math Works Inc., Natick,
MA, 2002.
[8] S. Li and R. Challoo, Restructuring
electric machinery course with an
integrative approach and computer-
assisted teaching methodology, IEEE
Trans Educ. 49 (2006), 16_28.
[9] J. J. D‟Azzo and C. H. Houpis, Linear
control system analysis and design,
McGraw-Hill, New York, 1995.
[10] S. J. Chapman, Electric machinery
fundamentals, 3rd
ed., WCB/McGraw-Hill,
New York, 1998.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-1
A SURVEY ON CURRENT CONVEYOR: NOVEL
UNIVERSAL ACTIVE BLOCK
1 Indu Prabha Singh and 2 Dr. Kalyan Singh
1Deptt. of Electronics and Comm. Engg. SITM, UNNAO-209859, India
2Dept. of Physics and Electronics Engg. , Dr. RML AVADH UNIVERSITY, FAIZABAD, India
(Email: [email protected], [email protected] )
ABSTRACT
Current conveyors (CCs) are being
increasingly employed to replace operational
amplifiers in almost all analog signal-
processing applications because their current
mode architectures are particularly suitable
for today’s low-voltage high frequency
applications. In this paper, different basic
current amplifier topologies are discussed.
With these amplifiers, more complicated
current-mode amplifiers are constructed in the
CMOS Integration technology.
Key Words: current conveyor, current
feedback amplifier, operational amplifier,
higher-bandwidth, CCI.
1. INTRODUCTION
Current conveyors have been used as a
basic building block in a variety of
electronic circuit in instrumentation and
communication systems. Today they
replace the conventional Op-amp in so
many applications such as active filters,
analog signal processing, and converters.
The current-conveyor, published in 1968
[1], represented the first building block
intended for current signal processing. In
1970 appeared the enhanced version of the
current-conveyor: the second-generation
current-conveyor CCII [2]. They used high
quality PNP - NPN transistors of a like
polarity and match each other but
difference in current gain reduced the
circuit accuracy. This is due to the base
current error. The other current conveyor
was devised in 1984 by G. Wilson [3]
where another current-mirror configuration;
known as Wilson current mirror was
employed. It consists of an operational
amplifier and external PNP transistors. A
second generation current conveyor (CCII)
was presented in 1990 using an operational
amplifier and external CMOS transistors
[4]. Both circuits were subject to the OP-
AMP performance and again to the
transistor mismatching. During that time,
research societies started to notice that the
voltage-mode operational amplifier is not
necessarily the best solution to all
analogue circuit design problems. New
research findings regarding current-mode
signal processing using current-conveyors
were presented. Furthermore, a
commercial product became available: the
current-feedback operational amplifier [5,
6]. The high slew rate and wide bandwidth
of this amplifier resulted in its popularity
in video amplifier applications.
The current conveyor is receiving
considerable attention as they offer analog
designers some significant advantages over
the conventional op-amp. These
advantages can be pointed out as follows:
Improve AC performance with better
linearity.
Wider and nearly constant bandwidth
independent of closed loop gain.
Relatively High slew rate (typically
2000V/μs).
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-2
Flexibility of driving current or voltage
signal output at its two separate nodes,
hence suitable for current and voltage mode
devices.
Reduced supply voltage of integrated
circuits.
Accurate port transfer ratios equal to
unity hence employed in low sensitivity
design.
Requirement of smaller number of
passive components to perform a specific
function.
In 1988 the principle of a MOS current
copier was presented [7], which enabled
analogue circuit designers to design
different Current Conveyors using only
MOS-transistors. Therefore, apart from
above advantages following are the driving
force behind the development of MOS
Current Conveyor:
Analog VLSI addresses almost all real
world problems and finds exciting new
information processing applications in
variety of areas such as integrated
sensors, image processing, speech
recognition, hand writing recognition
etc [5]. The need for low-voltage low-
power circuits is immense in portable
electronic equipments like laptop
computers, pace makers, cell phones etc.
Voltage Mode Circuits are rarely used
in low-voltage circuits as the minimum
bias voltages depend on the threshold
voltages of the MOSFETs. However, in
current mode circuits (CMCs), the
currents decide the circuit operation and
enable the design of the systems that
can operate over wide dynamic range.
MOS-transistors in particular are more
suitable for processing currents rather
than voltages because the output signal
is current both in common-source and
common-gate amplifier configurations.
Common-drain amplifier configuration
is almost useless at low supply voltages
because of the bulk-effect present in
typical CMOS-processes.
MOS current-mirrors are more
accurate and less sensitive to process
variation. Therefore, MOS-transistor
circuits should be simplified by using
current signals in preference to voltage
signals. For this reason, integrated
current-mode system realizations are
closer to the transistor level than the
conventional voltage-mode realizations.
When signals are widely distributed as
voltages, the parasitic capacitances are
charged and discharged with the full
voltage swing, which limits the speed
and increases the power consumption
of voltage-mode circuits. Current-
mode circuits cannot avoid nodes with
high voltage swing either but these are
usually local nodes with less parasitic
capacitances. Therefore, it is possible
to reach higher speed and lower
dynamic power consumption with
current-mode circuit techniques.
Current-mode interconnection circuits
in particular show promising
performance
2. BASIC CURRENT CONVEYOR
A CC is a three or more port (X, Y, Z)
network. Whose input-output relationship
is given by:
(1)
Where A, B, C assume a value either 1, 0
or - 1 and RX is the intrinsic resistance
offered by the port X to the input currents.
For an ideal CC VX = VY and the input
resistance (RX) at port X is zero (equation
(1)). But in practical CCs, RX is a nonzero
positive value.
The commonly used block representation
of a CC is shown in Figure 1, where X and
Y are the input terminals and Z is the
output terminal.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-3
Figure1: GeneralCurrentConveyor Symbol
3. FIRST GENERATION CURRENT-
CONVEYOR (CCI)
The first generation current conveyor CCI
forces both the currents and the voltages in
ports X and Y to be equal and a replica of
the currents is mirrored (or conveyed) to
the output port Z. Port Y is used as input
for voltage signals and it should not load
the input voltage source by drawing
current. But, in some applications, it is
desirable to draw currents from the input
voltage source. When A = 1, port Y draws
a current equal to the current injected at
port X and the configuration is termed as
CCI.
Figure 2 presents a simple MOS
implementation of the first generation
current-conveyor CCI. In this circuit, the
NMOS transistors M1 and M2 form a
current mirror that forces the drain currents
of the PMOS transistors M3 and M4 to be
equal and hence the voltages at the
terminals X and Y are forced to be
identical.
Figure2: simple MOS implementation of the
first generation current-conveyor CCI
Because of this low impedance at the input
terminal CCI circuit can be used as an
accurate current amplifier. In addition, the
DC-voltage level at the current input X can
be easily set to a desired value by the
voltage at the Y-terminal and input
voltage-to-current conversion is easier .It
can also be used as a negative impedance
converter (NIC) [12], if the Y-terminal is
terminated with a grounded resistance R.
The impedance at the terminal X equals
4.SECOND GENERATION CURRENT-
CONVEYOR ( CCII)
In many applications, only one of the
virtual grounds in terminals X and Y of the
first generation current-conveyor is used
and the unused terminal must be grounded
or otherwise connected to a suitable
potential. This grounding must be done
carefully since a poorly grounded input
terminal may cause unwanted negative
impedance at the other input terminal.
Moreover, for many applications a high
impedance input terminal is preferable. For
these reasons, the second generation
current-conveyor was developed. It has
one high and one low impedance input
rather than the two low impedance inputs
of the CCI [2, 18-20].
Figure3: The principle of the second
generation current-conveyors.
(a) The positive conveyor CCII+, iz = ix.
(b) The negative conveyor CCII-, iz =- ix.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-4
This current-conveyor differs from the first
generation conveyor in that the terminal Y
is a high impedance port, i.e. there is no
current flowing into Y(A = 0). The Y-
terminal of the second generation current-
conveyor is a voltage input and the Z-
terminal is a current output, the X-terminal
can be used both as a voltage output and as
a current input. Therefore, this conveyor
can easily be used to process both current
and voltage signals unlike the first
generation current-conveyor or the
operational amplifier.
A further enhancement to the second
generation current-conveyor is that there
are two types of conveyors: in the positive
current-conveyor CCII+, the currents ix
and iz have the same direction as in a
current-mirror and in the negative current-
conveyor CCII-the currents ix and iz have
opposite direction as in a current buffer.
The second-generation current-conveyor is
in principle a voltage-follower with a
voltage input, Y, and a voltage output, X,
and a current-follower (or current-inverter)
with a current input X and a current output
Z connected together (Figure 3). The
negative second-generation current-
conveyor CCII- can also be considered an
idealised MOS-transistor, where the
currents iy =ig =0 and iz =id = - ix = - is and
the voltages vx = vs = vy = vg. An ideal
MOS transistor is one that has a zero
threshold voltage Vt and zero channel
length modulation parameter ϒ and
operates in the saturation region regardless
of the drain-source voltage (positive or
negative).
5. THIRD GENERATION CURRENT-
CONVEYOR (CCIII)
Current-conveyor III was proposed in
1995 [9]. The operation of the third
generation current-conveyor CCIII is
similar to that of the first order current-
conveyor CCI, with the exception that the
currents in ports X and Y flow in opposite
directions (A= -1). As the input current
flows into the Y-terminal and out from the
X-terminal, the CCIII has high input impedance with common-mode current signals, i.e. identical currents are fed both to Y- and X-terminals. Therefore common-mode currents can push the input terminals out from the proper operation range. Therefore this conveyor is used as current probing.
6. CURRENT-FEEDBACK OPERAT-
IONAL AMPLIFIER
The current-feedback operational amplifier
is positive second generation current-
conveyor CCII+ with an additional voltage
buffer at the conveyor current output (5, 7,
8). The non-inverting port (Y) exhibits high
impedance to voltage signals where as the
inverting port (X) present low impedance to
the input current signals. The current at the
inverting input (X) of the current-feedback
operational amplifier is transferred to the
high impedance current-conveyor output Z,
causing a large change in output voltage.
The current-feedback operational amplifier
has a transresistance equal to the
impedance level at the conveyor Z-output.
Therefore, in the literature, the current-
feedback operational amplifier is also
referred to as a transimpedance amplifier.
The most commercial current-feedback
operational amplifier is AD844 [11],
where the user has access to the high
impedance node TZ. This amplifier can
also be utilised as a second generation
current-conveyor and current to-voltage
converter. The applications and advantages
Figure 4: The operating principle of the
current-feedback operational amplifier.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-5
in realizing active filter transfer function
using CFAs have received great attention
because the amplifier enjoys the feature of
constant feedback independent of closed
loop gain and high slew rate besides
having low output impedance. Thus it is
advantageous to use CFA as a basic
building block in the accomplishment of
various analog signal-processing tasks.
7. OPERATIONAL FLOATING
CONVEYOR The operational floating conveyor (Figure 5) is a current-mode building block that combines the transmission properties of a current-conveyor and a current-feedback operational amplifier, and has an additional output current sensing capability [13]. The matrix representation of the operational floating conveyor is
(2)
Where Zt is the transimpedance of the
internal current-feedback operational
amplifier.
If a current-conveyor is a voltage-follower
with an additional output current-sensing
circuit, the operational floating conveyor is
a current-feedback operational amplifier
with a similar output current-sensing
circuit. Alternatively this conveyor can be
constructed of two cascaded current-
Figure 5: The operational floating conveyor
constructed of two second generation current-
conveyors.
conveyors. With this circuit, it is possible
to realise all four types of amplifiers:
voltage, current, transconductance, and
transimpedance amplifiers, as presented in
Figure 6. The voltage amplifier in Figure 6
operates identically to the current-feedback
operational amplifier realization of the
noninverting voltage amplifier.The four
amplifier types can also be realised with
second generation current-conveyors as
open loop amplifiers. However, when
operational floating conveyor realisations
are used, the amplifier gain is less sensitive
to finite X-terminal impedance. Since the
feedback reduces impedance levels at both
X- and W-terminals, the band- widths of
the amplifiers are less sensitive to parasitic
capacitances. Furthermore, the feedback
also reduces distortion at low frequencies
but still the current signal path from W- to
Z-terminal remain outside the feedback
loop and thus the nonlinearity remains
unchanged in that part.
8. COMPOSITE CONVEYORS
The operational floating conveyor can be
also configured to form a high
performance second generation current-
conveyor as presented in Figure 7a. This is
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-6
Figure 6: Basic amplifier types realised with
operational floating conveyor.
(a) Voltage amplifier (b) Current amplifier
(c) Transconductance (d) Transresistance
a useful technique for designing CMOS
current-conveyors: with two poorly
operating simple CMOS positive second
generation conveyors, one positive
conveyor with enhanced X-terminal
impedance Zx can be constructed. In the
case of simple CMOS conveyors even the
resistor RF can generally be omitted as the
X-terminal is high enough to prevent any
stability and settling problems. There is an
alternative way to construct a composite
conveyor which lowers the X-terminal
impedance. This composite CCII- is
presented in Figure 7 b [14]. In this
composite conveyor, the lower conveyor
CC2 works as a negative impedance
conveyor and consequently the X-terminal
impedance of the composite CCII- is
Figure 7: Different composite conveyors.
(a) A composite CCII+ with enhanced Zx
resembling an operational floating
conveyor.
(b) A composite CCII- with a different
technique to lower Zx.
(c) A composite CCII+ with enhanced Yx.
9.FULLY DIFFERENTIAL CURRENT
CONVEYOR ( FDCCII) Recently, a new active element called the
FDCCII has been proposed [15] to
improve the dynamic range in mixed-
mode applications where fully differential
signal processing is required. The matrix
input–output relationship of the eight-
terminal FDCCII is:
(3)
10. OPERATIONAL FLOATING
CURRENT CONVEYOR (OFCC)
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-7
The OFCC is a five-port network,
comprised of two inputs and three output
ports, as shown in matrix representation 4.
In this representation, the port labelled X
represents a low-impedance current input,
port Y is a high-impedance input voltage,
W is a low-impedance output voltage, and
Z+, and Z- are the high-impedance current
outputs with opposite polarities. The
OFCC operates where the input current at
port X is multiplied by the open loop
transimpedance gain to produce an output
voltage at port W. The input voltage at
port Y appears at port X and, thus, a
voltage tracking property exists at the
input port. Output current flowing at port
W is conveyed in phase to port Z+ and out
of phase with that flowing into port Z-, so
in this case, a current tracking action exists
at the output port. Thus, the transmission
properties of the ideal OFCC can be
conveniently described as
(4)
where iy and vy are the inward current and
voltage at the Y port, respectively, as
shown in Fig. 4. ix and vx are the input
current and voltage at the X port,
respectively. iw and vw are the output
current and voltage at W port, respectively.
iZ and vZ and are the output current and
voltage at Z+ port, respectively. Similarly,
iZ- and vZ- and are the output current and
voltage at the Z- port, respectively.
11. CONVERSION OF VOLTAGE-
MODE CIRCUIT TO CURRENT –
MODE: ADJOINT PRINCIPLE
As a wide range of voltage-mode analog
circuits already exist, a straight forward
method of converting these voltage-mode
circuits to current-mode circuits would be
very useful. In such a method a circuit
using voltage amplifiers and passive
components is converted into one that
contains current amplifiers and passive
components. An ideal voltage amplifiers
has infinite input impedance and zero
output impedance, while an ideal current
amplifier has zero input impedance and
infinite output impedance. Consequently,
direct replacement of a voltage amplifier
with a current amplifier will lead to
different circuit behaviour.
A voltage-mode circuit can be converted
into a current-mode circuit by constructing
an interreciprocal network by using the
adjoint principle [1, 17].According to this
principle, a network is replaced with an
adjoint network Na, the voltage excitation
is interchanged to a current response, and
the voltage response is interchanged to a
current excitation, as demonstrated in
Figure 8.Thus, the resulting transfer
functions of these two networks N and Na
are identical:
(5)
Figure 8: Interreciprocal networks N and Na.
The networks N and Na are thus said to be
inter-reciprocal to one another. When the
networks N and Na are identical, for
example in the case of passive networks,
the networks are said to be reciprocal. In
order to maintain identical transfer
functions for both the original network N
and the adjoint network Na the impedance
levels in the corresponding nodes of both
networks should be identical. Therefore,
the signal flow is reversed in the adjoint
network and a voltage source is converted
to a current sensing element as they both
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-8
behave as short circuits. Similarly, a
voltage sensing element is converted to a
current source. A list of circuit elements
and their adjoint elements are presented in
the Table 1.
Table 1: Some circuit elements with their
corresponding adjoint elements
In addition, controlled sources can be
converted with the same principles: the
signal flow is reversed and the impedance
level is kept the same. In this way, a
voltage amplifier is converted to a current
amplifier and a current amplifier is
converted to a voltage amplifier,
respectively. However, since
transresistance and transconductance
amplifiers are inter-reciprocal, networks
containing only transresistance or
transconductance amplifiers and passive
elements differ only in signal direction and
type.
The adjoint principle can also be applied to
transistor level circuits. In this case, a
bipolar transistor in a common-emitter
amplifier configuration is inter-reciprocal
to itself and the common-collector
amplifier configuration has the common-
base configuration as its adjoint.
Converting a voltage-mode bipolar
transistor circuit to a current-mode MOS-
transistor circuit could be beneficial as it
minimises the use of source-follower
stages which have poor low-voltage
performance due to the bulk effect. Bipolar
transistor circuits are conventionally
constructed of common-emitter and
common-collector amplifier stages and the
resulting MOS-transistor adjoint circuit is
constructed of common-source and
common-gate amplifier stages.
Figure 9: Sallen- Key active biquad Filter
using Op-amp
Figure10: Sallen-Key active biquad Filter
using Current conveyor II
Using the adjoint principle the low-pass
Sallen-Key circuit can be replaced with a
current conveyor based circuit. The
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-9
transfer function is the same for both of the
circuits
(a)
(b) Figure 11: Instrumentation amplifiers (a) using
Op-Amp (b) using CCII+
As shown in figure11, Instrumentation
amplifiers implemented with three op-
amps. It requires several matched resistors
to guarantee high CMRR because of the
limited gain-bandwidth product of the
high-gain amplifiers the bandwidth of the
CMRR is limited. A differential amplifier
with high CMRR can be also realised with
two current conveyor and two resistors
without any matching components. The
bandwidth of the current conveyor based
amplifier is large with high voltage gains
as current conveyors operate in open-loop
without the gain-bandwidth product
limitation.
12. CONCLUSION
In this paper, we have presented
classification and some advanced
applications of CCs. The concept of
modularity has been introduced in analog
circuit design through reconfiguring a
current conveyor as CFAs and OFCs.
Some of the performance parameters are
bandwidth, power dissipation etc. We have
seen that these CCs can perform better
than operational amplifier based circuits in
almost all signal-processing applications.
These circuits are now used in custom-
built analog ICs. It is possible to explore
new type of devices and their applications.
REFERENCES
[1] K. Smith, A. Sedra, “The current-conveyor-a new circuit building block,” IEEE Proc., vol. 56, pp. 1368-69, 1968. [2] A. Sedra, K. Smith, “A second-generation current-conveyor and its applications,” IEEE Trans., vol. CT-17, pp. 132-134, 1970. [3] D. Frey, “Log-domain filtering: an approach to current-mode filtering,” IEE Pro-ceedings G, vol. 140, pp. 406-416, Dec. 1993. [4] J. Hughes, N. Bird, I. Macbeth, “Switched currents - a new technique for analog sampled-data signal processing,” in Proc. IEEE Int. Symposium on Circuits and Systems (ISCAS’89), Portland, USA, 1989, pp. 1584-1587. [5] C. Toumazou, F. J. Lidgey, D. G. Haigh (ed.), Analogue IC design: the current-mode approach, London, Peter Peregnirus Ltd, 1990, 646 p. [6] Comlinear Corporation, A new approach to op-amp design, Application Note 300-1, March 1985. [7] S. Daubert, D. Vallancourt, Y. Tsividis, “Current copier cells,” Electronics Let-ters, vol. 24, pp. 1560-1561, Dec. 1988. [8] F. J. Lidgey, K. Hayatleh, “Current-feedback operational amplifiers and applica-tions,”Electronics &
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0302-10
Communications Engineering Journal, vol. 9, pp. 176-182, Aug. 1997. [9] A. Fabre, “Third-generation current conveyor: a new helpful active element,” Electronics Letters, vol. 31, pp. 338-339, March 1995. [10] K. Manetakis, C. Toumazou, “A new high-frequency very low output impedance CMOS buffer,” Proc. IEEE Int. Symposium on Circuits and Systems (ISCAS-96), Atlanta, 1996 pp. 485-488. [11] Soliman, A. M. (1998) A new filter
configuration using current feedback
op-amp. Micoelectronics Journal, 29, 409-419.
[12] R. Brennan, T. Viswanathan, J. Hanson, “The CMOS negative impedance con-verter,” IEEE J. Solid State Circuits, vol. 23, pp. 1272-1275, Oct. 1988. [13] C. Toumazou, A. Payne, F. Lidgey, “Operational floating conveyor,” Electronic Letters, vol. 27, pp. 651-652, April 1991. [14] A. Fabre, H. Barthelemy, “Composite second-generation current conveyor with reduced parasitic resistance,” Electronic Letters, vol. 30, pp. 377-378, March1994. [15] 1 El-Adawy, A.A., Soli an, A.M., and
Elwan, H.O.: „A novel fully differential
current conveyor and applications for analog
VLSI‟, IEEE Trans Circuits Syst. II, Analog
Digit. Signal Process, 2000, 47,(4),pp. 306–
[18] Sanchez-Sinencio, E., and Andreou,
A. G. ed. (1999). “Low voltage/low power
integrated circuits and systems”, IEEE
Press.
[19] Barthelemy, H., and Fabre, A. (2002).
“A second generation current-controlled
conveyor with negative intrinsic
resistance”, IEEE Trans. Circuits and
Systems-I, 49(1), 63–65.
[20] Fabre, A., and Alami, M. (1997). “A
precise macromodel for second generation
current-conveyors”, IEEE Trans. Circuits
and Systems-I, 44(7), 639–642.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0401-1
A COMPARITIVE STUDY OF GENETIC ALGORITHM
AND THE PARTICLE SWARM OPTIMIZATION
Sapna Katiyar, Deepika pandey, Sakshi Chhabra, Vaishali Gupta,
A.B.E.S. Institute of Technology, NH-24, Vijay Nagar, Ghaziabad (UP) 201009,
(e-mail: [email protected], [email protected], [email protected],
Corresponding Author: [email protected]
ABSTRACT
Particle Swarm Optimization (PSO) is a relatively recent heuristic search method whose mechanics are
inspired by the swarming or collaborative behavior of biological populations.PSO is similar to the Genetic
Algorithm (GA) as these two evolutionary heuristics are population-based search methods. In other words,
PSO and the GA move from a set of points (population) to another set of points in a single iteration with
likely improvement using a combination of deterministic and probabilistic rules. The GA and its many
versions have been popular in the academy and the industry mainly because of its intuitiveness, ease of
implementation, and the ability to effectively solve highly nonlinear, mixed integer optimization problems
that are typical of complex engineering systems. The drawback of the GA is its expensive computational cost.
This paper attempts to examine the claim that PSO has the same effectiveness (finding the true global optimal
solution) as the GA but with significantly better computational efficiency (less function evaluations) by
implementing statistical analysis and formal hypothesis testing. The major objective of this paper is to
compare the computational effectiveness and efficiency of the GA and PSO using a formal hypothesis testing
approach.
Index Terms - Genetic algorithm, Candid solution, metaheuristics, Numerical optimization, stochastic, swarm,
variants, hybrid PSO, PSO parameters
__________________________________________________________________________________________
I. BACKGROUND
The Genetic Algorithm (GA) was introduced in the
mid 1970s by John Holland and his colleagues and
students at the University of Michigan. The GA is
inspired by the principles of genetics and evolution,
and mimics the reproduction behavior observed in
biological populations. The GA employs the principal
of “survival of the fittest” in its search process to
select and generate individuals (design solutions) that
are adapted to their environment (design
objectives/constraints). Therefore, over a number of
generations (iterations), desirable traits (design
characteristics) will evolve and remain in the genome
composition of the population (set of design solutions
Generated each iteration) over traits with weaker
undesirable characteristics. The GA is applied to
solve complex design optimization problems because
it can handle both discrete and continuous variables
and nonlinear objective and constrain functions
without requiring gradient information.
Particle Swarm Optimization (PSO) was invented
by Kennedy and Eberhart in the mid 1990s while
attempting to simulate the choreographed, graceful
motion of swarms of birds as part of a socio cognitive
study investigating the notion of “collective
intelligence” in biological populations. In PSO, a set
of randomly generated solutions (initial swarm)
propagates in the design space towards the optimal
solution over a number of iterations (moves) based
on large amount of information about the design
space that is assimilated and shared by all members
Of the swarm. PSO is inspired by the ability of flocks
of birds, schools of fish, and herds of animals to
adapt to their environment, find rich sources of food,
and avoid predators by implementing an “information
sharing” approaches, hence, developing an
evolutionary advantage.
II. GENETIC ALGORITHM
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0401-2
Start
Create randomly selected
initial population
Fitness
Evaluation
Stop
Reproduction
Mutation
Discard
Fit Chromosome?
Optimal
Solution?
No
No
Yes
Yes
Fig 1: Flow Chart of the Genetic Algorithm
In a genetic algorithm, a population of strings
(called chromosomes or the genotype of the genome),
which encode candidate solutions (called phenotypes)
to an optimization problem, evolves toward better
solutions. Solutions are represented in binary as
strings of 0s and 1s, but other encodings are also
possible. The evolution usually starts from a
population of randomly generated individuals and
happens in generations. In each generation, the
fitness of every individual in the population is
evaluated, multiple individuals are stochastically
selected from the current population (based on their
fitness), and modified (recombined and possibly
randomly mutated) to form a new population. The
new population is then used in the next iteration of
the algorithm. The algorithm terminates when either
a maximum number of generations has been
produced, or a satisfactory fitness level has been
reached for the population. If the algorithm has
terminated due to a maximum number of generations,
a satisfactory solution may or may not have been
reached.
A typical genetic algorithm requires:
A genetic representation of the
solution domain
A fitness function to evaluate the solution
domain.
A standard representation of the solution is as
an array of bits. The main property that makes these
genetic representations convenient is that their parts
are easily aligned due to their fixed size, which
facilitates simple crossover operations. Variable
length representations may also be used, but
crossover implementation is more complex in this
case. Tree-like representations are explored in genetic
programming and graph-form representations are
explored in evolutionary programming.
A. OBJECTIVE FUNCTION OF GENETIC
ALGORITHM
The fitness function is defined over the
genetic representation and measures the quality of the
represented solution. The fitness function is always
problem dependent. For instance, in the knapsack
problem one wants to maximize the total value of
objects that can be put in a knapsack of some fixed
capacity.
A representation of a solution might be an
array of bits, where each bit represents a different
object, and the value of the bit (0 or 1) represents
whether or not the object is in the knapsack. Not
every such representation is valid, as the size of
objects may exceed the capacity of the knapsack.
The fitness of the solution is the sum of
values of all objects in the knapsack if the
representation is valid or 0 otherwise. In some
problems, it is hard or even impossible to define the
fitness expression; in these cases, interactive genetic
algorithms are used.
Once we have the genetic representation and
the fitness function defined, GA proceeds to initialize
a population of solutions randomly, and then improve
it through repetitive application of mutation,
crossover, and inversion and selection operators.
B. IMPLEMENTATION ALGORITHM
The genetic algorithm uses the chromosomes fitness
value to create a new population consisting of the
fittest members. The flow chart of the GA is given
here in Fig 1.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0401-3
C. APPLICATIONS
Genetic algorithms find application in bioinformatics,
phylogenetics, computational science, engineering,
economics, chemistry, manufacturing, mathematics,
physics and other fields.
Learning Robot behavior using Genetic Algorithms:
Robot has become such a prominent tools that it has
increasingly taken a more important role in many
different industries. As such, it has to operate with
great efficiency and accuracy. This may not sound
very difficult if the environment in which the robot
operates remain unchanged, since the behaviors of
the robot could be pre-programmed. However, if the
environment is ever-changing, it gets extremely
difficult, if not impossible, for programmers to figure
out every possible behaviors of the robot. Applying
robot in a changing environment is not only
inevitable in modern technology, but is also
becoming more frequent. This has obviously led to
the development of a learning robot.
III.PARTICLE SWARM OPTIMIZATION
The PSO was first designed to simulate birds seeking
food which is defined as a “cornfield vector.” The
bird would find food through social cooperation with
other birds around it (within its neighborhood). It was
then expanded to multidimensional search.
Particle swarm optimization (PSO) is a
computational method that optimizes a problem
by iteratively trying to improve a candidate solution
with regard to a given measure of quality. Such
methods are commonly known as metaheuristics as
they make few or no assumptions about the problem
being optimized and can search very large spaces of
candidate solutions. However, metaheuristics such as
PSO do not guarantee an optimal solution is ever
found.
PSO does not use the gradient of the problem being
optimized, which means PSO does not require for the
optimization problem to be differentiable as is
required by classic optimization methods such
as gradient descent and quasi-newton methods. PSO
can therefore also be used on optimization problems
that are partially irregular, noisy, change over time,
etc.
PSO optimizes a problem by having a population
of candidate solutions, here dubbed particles, and
moving these particles around in the search-space
according to simple mathematical formulae. The
movements of the particles are guided by the best
found positions in the search-space which are
updated as better positions are found by the particles.
PSO algorithm works by having a population (called
a swarm) of candidate solutions (called particles).
These particles are moved around in the search-space
according to a few simple formulae. The movements
of the particles are guided by their own best known
position in the search-space as well as the entire
swarm's best known position. When improved
positions are being discovered these will then it will
guide the movements of the swarm. The process is
repeated and satisfactory solution will be discovered.
A.PSO VARIANTS
Various variants of a basic PSO algorithm are
possible. New and some more sophisticated PSO
variants are continually being introduced in an
attempt to improve optimization performance. There
is a trend in that research; one can make a hybrid
optimization method using PSO combined with other
optimization techniques.
Discrete PSO
Constriction Coefficient
Bare-bones PSO
Fully informed PSO
B. APPLICATIONS
The first practical application of PSO was in the field
of neural network training and was reported together
with the algorithm itself (Kennedy and Eberhart
1995). Many more areas of application have been
explored ever since, including telecommunications,
control, data mining, design, combinatorial
optimization, power systems, signal processing, and
many others. PSO algorithms have been developed to
solve:
Constrained optimization problems
Min-max problems
Multi objective optimization problems
Dynamic tracking
C. IMPLEMENTATION ALGORITHM
The PSO algorithm is simple in concept, easy to
implement and computational efficient. Original PSO
was implemented in a synchronous manner (Fig 2)
but improved convergence rate is achieved by
asynchronous PSO algorithm Fig 2.
CONFERENCE ON “SIGNAL PROCESSING AND REAL TIME OPERATING SYSTEM (SPRTOS)” MARCH 26-27 2011
COS0401-4
Fig 2: Synchronous PSO algorithm (Parallel
Processing)
IV. GENETIC ALGORITHM VERSUS
PARTICLE SWARM OPTIMIZATION
1) GA is inherently discrete, i.e. it encodes the
design variables into bits of 0’s and 1’s,
therefore it easily handles discrete design
variables, and PSO is inherently continuous
and must be modified to handle discrete
design variables.
2) Unlike the GA with its binary encoding, in
PSO, the design variables can take any
values, even outside their side constraints,
based on their current position in the design
space and the calculated velocity vector.
V. CONCLUSIONS
Particle Swarm Optimization (PSO) is a relatively
recent heuristic search method that is based on the
idea of collaborative behavior and swarming in
biological populations. PSO is similar to the Genetic
Algorithm (GA) in the sense that they are both
population-based search approaches and that they
both depend on information sharing among their
population members to enhance their search
processes using a combination of deterministic and
probabilistic rules. Conversely, the GA is a well
established algorithm with many versions and many
applications.
GA is very helpful when the developer does not have
precise domain expertise, because GAs possesses the
ability to explore and learn from their domain. PSO
can be applied to multi-objective problems, in which
the fitness comparison takes pareto dominance into
account when moving the PSO particles and non-
dominated solutions are stored so as to approximate
the pareto front.
The objective of this research paper is to test
the hypothesis that states that although PSO and the
GA on average yield the same effectiveness (solution
quality), PSO is more computationally efficient (uses
less number of function evaluations) than the GA.
REFERENCES
[1] Goldberg, D.E. (1989). Genetic Algorithms in
Search, Optimization and Machine Learning,
Addison-Wesley.
[2] Kennedy, J. and Eberhart, R., “Particle Swarm
Optimization,” Proceedings of the IEEE
International Conference on Neural Networks,
Perth, Australia 1995, pp. 1942-1945.
[3] Kennedy, J. and Eberhart, R., Swarm Intelligence,
Academic Press, 1st ed., San Diego, CA, 2001.
[4] Yuhui Shi. Particle Swarm Optimization.
Electronic Data Systems, Inc. IEEE Neural Network
Society.
[5] R. Poli. Analysis of the publications on the
applications of particle swarm optimization. Journal
of Artificial Evolution and Applications, Article ID
685175, 10 pages, 2008.
[6] R. Poli, J. Kennedy, and T. Blackwell. Particle
swarm optimization. An overview. Swarm
Intelligence, 1(1):33-57, 2007.