International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 16 (2016) pp 9099-9105
© Research India Publications. http://www.ripublication.com
9099
Innovative improved Direct Torque Control of Doubly Fed Induction
Machine (DFIM) using Artificial Neural Network (ANN-DTC)
Abderrahim ZEMMIT 1, Sabir MESSALTI 1 , Abdelghani HARRAG1,2
1Electrical Engineering Department, Faculty of Technology, Mohamed Boudiaf University, 28000 Msila , Algeria. 2CCNS Laboratory, Electronics Department, Faculty of Technology, Ferhat Abbas University, 19000 Setif, Algeria.
Abstract
This paper presents a new advanced Direct Torque Control of
doubly fed induction motor using Artificial Neural Network
controllers, which conventional IP and switching table have
been replaced by new Artificial Neural Network controllers to
overcome the most drawbacks and limitations of classical
DTC control especially high torque and flux ripples, accuracy
and the convergence speed under sudden changing torque
reference and/or speed reference. The Artificial Neural
Network controllers parameters that optimizes the
performances of conventional Direct Torque Control of
doubly fed induction motor have been optimized in offline
mode where several training sets have been tested based on
conventional IP and switching table rules, then the best
Artificial Neural Network controllers are embedded in DTC-
DFIM system in online mode. The proposed ANN controllers
have shown better performance compared to conventional
DTC in both the transient and steady states, in which
numerous benefits related to flux and torque ripples
overshoot and response time have been confirmed.
Keywords: Artificial Neural Network, Double Feed Induction
Machine (DFIM), Direct Torque Control, DTC, ANN
Switching table, low torque and flux ripples.
INTRODUCTION
Today, due to huge progress in power electronics, doubly fed
induction machines (DFIMs) became one of the best
promising solutions for many applications especially for wind
energy conversion, variable speed application, railway
traction, marine propulsion, and hydroelectric power stations,
etc[1-3]. The widespread use of doubly fed induction
machines is justified by following advantages:
DFIMs can generate reactive current and produce
constant-frequency electric power at variable speed
operation.
DFIM can be fed and controlled stator or rotor by various
possible combinations;
The power flow can be modulated
DFIM can be operate in wide range of speed variation
around the synchronous speed (until ± 30%), so, it The
active and reactive power can be control separately;
the stator and rotor currents are measurable;
Many DFIMs parameters can be controlled independently
(torque, flux, and power factor).
Although, the previous advantages of DFIM, its brushes and
slip rings structure are the most drawbacks, which it require
permanent maintenance.
The growing interest of the doubly fed induction machine
(DFIM) has motivates many researchers to develop and
improve it continuously, which they study the including
different parts such as converter parts, conception machine,
materials and the control of DFIM. In last point, a large
number of techniques have been proposed since it can exhibits
many merits, the field oriented control (FOC) techniques are
developed firstly, However, it highly dependent on the
parameters of the induction machine and constant gains PI
controllers may become unable to provide the required control
performance. Hence PI controller limitations and its
complexity of implementation are the most drawbacks [4-9].
Moreover, Direct torque control (DTC) was proposed in
1980s and then it was well developed in power electronics and
drives application for its excellent steady state and transient
performance [10], and Direct Self Control (DSC) by
Depenbrock [11] have provided better steadystate and
transient torque control conditions rather than FOC
techniques. In addition, Direct Control techniques do not
require current regulators, nor coordinate transformations
or specific modulations like PWM or SVM for pulse
generation. However Direct Control techniques present some
many drawbacks compared to the FOC such as difficulties of
torque control at very low speeds, variable switching
frequency, and the lack of direct current control
In recent years , many improvements have introduced to
Direct Control techniques to reduce torque ripple [12-14] and
achieved constant switching frequency such as Direct
Mean Torque Control (DMTC) [7], Direct Torque
Control with Space Vector Modulation (DTC-SVM) [15]. In
addition, many intelligent control strategies have been
proposed such as artificial neural network[16-20], the fuzzy
logic control (FLC) [17], sliding mode control (SMC) [21-23],
backstepping control , Fuzzy-SVM , adaptive fuzzy vector
controller (AFVC) [17-28].
In this paper, new combined DTC-ANN method applied to
doubly fed induction motor has been proposed and studied.
Modelling of DFIM and details of proposed control strategy
have been presented. The performances in terms of torque
tracking, torque and flux ripples, and accuracy have been
demonstrated and compared.
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 16 (2016) pp 9099-9105
© Research India Publications. http://www.ripublication.com
9100
DOUBLY FED INDUCTION MACHINE MODEL
The state-all-flux DFIM dynamic model expressed in ( , )
axes rotational reference frame is given by the following
equations:
(1)
s s s s
s s s s
r r r r r
r r r r r
dV R IdtdV R IdtdV R IdtdV R Idt
(2)
s s s r
s s s r
r r r s
r r r s
l I MIl I MIl I MIl I MI
The electromagnetic torque is given by :
3
) 3( 2
em s r s rs
pMC I IL
DIRECT TORQUE CONTROL OF DFIM
DTC is an advanced control method involving direct control
of the electromagnetic torque and the flux developed in the
double feed Induction machine. At the heart of the control
system in the DTC are the flux and torque hysteresis
controllers and an optimal switching state logic block. An
appropriate DFIM model is essential for the correct estimation
of the electromagnetic torque and stator flux. The estimation
of these quantities is carried out by measurements of the
DFIM currents, flux linkages and the DC voltage. The
estimated values of the motor torque and flux are input to the
two hysteresis controllers wherein, a comparison between the
estimated and the actual values of the quantitates is performed
[25].
Direct torque control is based on the flux orientation, using
the instantaneous values of voltage vector. An inverter
provides eight voltage vectors, among which two are zeros.
These vectors are errors as well as the stator flux vector
chosen from a switching table according to the flux and torque
position. In this technique, we don’t need the rotor position in
order to choose the voltage vector. This particularity defines
the DTC as an adapted control technique of AC machines and
is inherently a motion sensorless control method [24-26]. The Fig.1 shows the schematic of Direct Torque Control of
doubly fed induction motor.
Figure 1: The diagram of the DFIM-DTC system.
As shown in Fig. 2, the position vectors and trajectory of the
stator flux is divided into six sectors. There are also 8 voltage
vectors which correspond to possible inverter states.
Figure 2: Vectors and Trajectory of the stator flux.
The flux estimator can be obtained by the following equation:
t
ssss dtIRVt0
) ( )( (4)
The stator flux Øsref and the torque Cemref are compared with
respective estimated values, and errors are processed through
hysteresis-band controllers. The digital outputs from the
hysteresis comparators along with the sector number are
shown in Table 1. The correct voltage vector is then selected.
The corresponding switch position for the inverter, to achieve
the selected voltage vector is shown in Table 2.
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 16 (2016) pp 9099-9105
© Research India Publications. http://www.ripublication.com
9101
Table 1: Switching table for voltage vectors.
Table 2: Switch positions and their voltage vectors.
The choice of the zero vectors (V0, V7) produces a smaller
torque and flux variations compared with the active vectors.
Then, the zero vectors are not really needed to keep the torque
and flux controlled; however, it is used to reduce the torque
and flux ripples at steady state operation. For almost every
application of DTC, it is advantageous if the torque and flux
ripples are minimized as much as possible [27].
PROPOSED DTC-ANN
The proposed DTC-ANN consist on two ANN controllers, the
first one replace the switching table which it providing the
voltage vector, the input of ANN controller are Cflux the stator
flux error , Ccpl the torque error and N sector number , this
ANN is based on feed -forward back propagation with four
hidden layers having 4,14,16 and 3 neurons in each layer
respectively and a logsig as activation functions, the output
layer has three neurons providing voltage vector, the proposed
ANN Switching Table is shown in Fig. 3.
Figure 3: The neural networks Switching Table with
Matlab /Simulink
The second ANN controller replace the conventional IP, it
based on feed forward back propagation with three hidden
layers having 10,14 and 1 neurons in each layer respectively
and a logsig as activation functions for first and second layer ,
however and the third layer has purelin activation function,
the input of ANN controller is the difference between
measured speed Wm and reference speed Wref , however the
output layer is Ceref . The proposed ANN Speed controller is
shown in Fig. 4. The weights are determined by training
algorithm.
Figure 4: The Neural Networks Speed Controller.
The proposed DTC-ANN scheme including both ANN
controllers is illustrated in the Fig.5.
Figure 5: DTC-ANN scheme .
SIMULATION RESULTS AND COMPARISON To demonstrate the efficiency of the proposed DTC-ANN
strategy, a comparative study between the conventional DTC
and the proposed DTC-ANN has been carried out using
0.8kW DFIM [4], which many performance parameters have
been studied.
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 16 (2016) pp 9099-9105
© Research India Publications. http://www.ripublication.com
9102
a) Zoom A
b) Zoom B
Figure 6: DTC-ANN under speed variation.
The simulation results obtained with no load as starting up
condition and connecting the nominal load as normal
operating condition are presented in following figures.
Figure 7: DTC-ANN under variable load.
The Fig.6 and Fig.7 illustrate the speed, electromagnetic
torque respectively, while Fig.8 shows the flux,
0.5 1 1.5 2 2.50
20
40
60
80
100
120
140
160
180
t(s)
Speed(rad/s
)
DTC-IP
DTC-ANN
Wref
Zoom A Zoom B
0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65
140
150
160
170
180
190
t(s)
Speed(rad/s
)
DTC-IP
DTC-ANN
Wref
IP
ANN
Response time :IP : 0.35sANN : 0.01 sReduction ratio : 1/35
Response time :IP : 0.24sANN : 0.62 sReduction ratio : 1/2.58
Zoom A
0.8 0.85 0.9 0.95 1 1.05
150
151
152
153
154
155
156
157
158
159
160
t(s)
Speed(r
ad/s
)
DTC-IP
DTC-ANN
Wref
Zoom B
ANNIP
Overshoot :IP : 7 rad/sANN : 0.8rad/s Reduction ratio : 1/8.75
0 0.5 1 1.5 2 2.5-4
-2
0
2
4
6
8
t(s)
Torq
ue(N
m)
DTC-IP
DTC-ANN
Cr
Zoom A
Zoom B
0.1 0.2 0.3 0.4 0.5 0.6-3
-2
-1
0
1
2
3
4
5
6
7
t(s)
Torq
ue(N
m)
DTC-IP
DTC-ANN
Cr
Zoom A
ANN
IP
Response time :IP : 0.57sANN : 0.25 sReduction ratio : 1/2.28
0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
-2
-1.5
-1
-0.5
0
0.5
t(s)
Torque(N
m)
DTC-IP
DTC-ANN
Cr
ANN IP Overshoot :IP : 1.78 NmANN : 0.11 NmReduction ratio : 1/16.18
Zoom A1
0.8 0.85 0.9 0.95 1 1.05
2.6
2.8
3
3.2
3.4
3.6
t(s)
Torq
ue(N
m)
DTC-IP
DTC-ANN
Cr
Zoom B
IPANNRipple :IP : 0.42 NmANN : 0.2 NmReduction ratio : 1/2.1
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 16 (2016) pp 9099-9105
© Research India Publications. http://www.ripublication.com
9103
From Fig. 6, we can see that at starting up with no load or in
case of nominal load, the DTC-ANN controller reaches its
speed reference rapidly without overshoot compared to the
conventional DTC. As consequence, the excellent dynamic
performance of torque and flux control is evident.
From Fig. 7, its clear that the electromagnetic torque obtained
by proposed DTC-ANN controller better performances
characterised by less response time and low overshoot.
The improvement of DTC-ANN controller regarding stator
flux ripple in steady state is undeniably clear in Fig. 8.
Figure 8: Stator flux reduction ripple.
Fig. 9. shows the flux signals issued by conventional DTC
and proposed DTC-ANN controllers, more than 47%
(0.055Wb for conventional DTC instead of 0.029Wb for
proposed DTC-ANN).
Figure 9: Nominal DFIM flux using DTC-ANN.
Table 3 summarizes the main improvements of the proposed
DTC-ANN compared to conventional DTC.
Table. 3. Comparative study between conventional DTC
and proposed DTC-ANN.
Performance DTC - IP DTC -
ANN
Improve-
ment (%)
W (rad/s) Response time (s) 0.62 0.24 61
Ouvershoot (rad/s) 7 0.8 88
Ce (N.m)
Response time (s) 0.57 0.25 56
Ouvershoot (N.m) 1.78 0.11 93.8
Ripple (N.m) 0.42 0.2 52.3
s (Wb)
Ripple (Wb) 0.055 0.029 47.2
IP ANN00
IP
V - VImprovement( ) = ×100
V
From Figures 6 to 9, and the Table 3, the proposed ANN-DTC
strategy constitutes a practical alternative to the conventional
control strategy and it has many features and advantages such
as:
a. It has good performances;
b. Reduced torque and flux ripples.
c. The ability to track the electromagnetic torque
driving the speed the its reference efficiently.
d. Fast flux and torque responses;
CONCLUSION In this paper, a novel improved direct torque control strategy
using ANN for DFIM (DTC-ANN) has been proposed and
investigated. In which, modelling of DFIM and proposed
DTC-ANN strategy have been discussed in detail. In addition,
the development of Artificial Neural Network controllers has
been explained (structure, training, layers, rules). A
comparative study between conventional DTC scheme and the
proposed DTC-ANN has been presented. Simulation results
demonstrate the superiority of the proposed DTC-ANN
compared to conventional DTC in term of faster converging
speed, better accuracy, and low ripple. Therefore the proposed
DTC-ANN is more suitable for DFIM control.
-1.5 -1 -0.5 0 0.5 1 1.5-1.5
-1
-0.5
0
0.5
1
1.5
Stator flux-a
Sta
tor
flux-b
DTC-IP
DTC-ANN
Zoom A
-1.06 -1.04 -1.02 -1 -0.98 -0.96 -0.94 -0.92 -0.9 -0.88 -0.86
-0.8
-0.75
-0.7
-0.65
-0.6
-0.55
Stator flux-a
Sta
tor
flux-b
DTC-IP
DTC-ANNZoom A
IPANN
0 0.5 1 1.5 2 2.50
0.2
0.4
0.6
0.8
1
1.2
1.4
t(s)
Sta
tor
Flu
x (
Wb)
DTC-IP
DTC-ANN
Zoom A
1.9 1.92 1.94 1.96 1.98 21.13
1.14
1.15
1.16
1.17
1.18
1.19
1.2
1.21
1.22
t(s)
Sta
tor
Flu
x (
Wb)
DTC-IP
DTC-ANN
Zoom A
ANN
IPRipple :IP : 0.055WbANN : 0.029 WbReduction ratio : 1/1.9
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 16 (2016) pp 9099-9105
© Research India Publications. http://www.ripublication.com
9104
APPENDIX:
Double Feed Induction Machine parameters:
Pn = 0.8 kW
Un = 220/380 V
F = 50 Hz
I = 3.8/2.2 A
Vr = 3×120 V; 4.1 A
Ω = 1420 tr/min
Rs = 11.98 Ω
Rr = 0.904 Ω
Ls = 0.414 H
Lr = 0.0556 H
M = 0.126 H
P = 2
J = 0.01 kg.m2
f = 0.001
REFERENCES
[1] Kianinezhad R., Nahid B., Betin F., Capolino G., A
novel Direct Torque Control (DTC) method for dual
three phase induction motors, IEEE, 2006, p. 939-
943.
[2] R. Cardenas, R. Pena ,“MRAS Observers for Sensor
less Control of Doubly Fed Induction Generators”,
IEEE Transactions on Power Electronics, Vol.23,
No.3, May 2008.
[3] S. Messalti, A. Gherbi, and S. Belkhiat,”
Improvement of Power System Transient of Stability
using wind farms based on doubly fed induction
generator (DFIG)”, Journal of Electrical Engineering
: Volume 13 ( 3), 2013
[4] Abderrahim ZEMMIT, Sabir MESSALTI,
“Modeling and Simulation of Doubly Fed Induction
Motor (DFIM) Control using DTC and DFOC: A
comparative study”, International Journal of Applied
Engineering Research ISSN 0973-4562 Volume 11,
Number 8 (2016) pp 5623-5628.
[5] Chikhi, M. Djarallah and K. Chikhil, “A
Comparative Study of Field-Oriented Control and
Direct-Torque Control of Induction Motors Using An
Adaptive Flux Observer”, Serbian Journal of
Electrical Engineeringvol. 7, No. 1, May 2010, 41-55
[6] D. Casadei, F. Profumo, G. Serra and A. Tani, “FOC
and DTC: two viable schemes for induction motors
torque control,” IEEE Trans. Power Electron., vol.
17, no. 5, pp. 779-787, Sept. 2002.
[7] E. Flach, R. Hoffmann, P. Mutschler, “Direct
mean torque control of an induction motor,” in
Proc. EPE’97 Conf., 1997.
[8] H. Merabet Boulouiha , A. Allali “Direct torque
control of multilevel SVPWM inverter in variable
speed SCIG-based wind energy conversion system,"
Science Direct ,Renewable Energy 80 (2015) 140-
152.
[9] F. Blaschke, “A new method for the structural
decoupling of A.C. induction machines,” in Conf.
Rec. IFAC, Dussesrdorf, Germany, Oct.1971.
[10] I. Takahashi and Y. Ohmori, “High-performance
direct torque control of an induction motor,” IEEE
Trans. Ind. Applicat., vol. 25, pp. 257-264, Mar./Apr.
1989.
[11] M. Depenbrock, ”Direct self-control (DSC) of
inverter-fed induction machine,” IEEE Trans.
Power Electron., vol. 3, pp. 420–429, Oct. 1988.
[12] J. K. Kang and S.K. Sul, “New direct torque
control of induction motor for minimum torque
ripple and constant switching frequency,” IEEE
Trans. Ind. Applicat., vol. 35, no.5, pp. 1076-1086,
Sept.-Oct. 1999.
[13] K. B. Lee, J-H. Song, I. Choy and J-Y. Yoo, “Torque
ripple reduction in DTC of induction motor driven by
three-level inverter with low switching frequency,”
IEEE Trans. Power Electron., vol. 17, no. 2, pp. 255-
264, March 2002.
[14] C.A. Martins, X. Roboam, T.A. Meynard, A.S.
Carvalho, “Switching frequency imposition and
ripple reduction in DTC drives by using a multilevel
converter,” IEEE Trans. Power Electron., vol. 17, no.
2, pp. 286–297, March 2002.
[15] Habetler, T.G.; Profumo, F.; Pastorelli, M.;
Tolbert, L.M.; “Direct torque control of induction
machines using space vector modulation”, IEEE
Trans. Ind. Applicat., vol. 28, no. 5, Sept.-Oct. 1992.
[16] S.M. Gadoue, D. Giaouris, J.W. Finch, “ Artificial
intelligence-based speed control of DTC induction
motor drives A comparative study ", Science Direct ,
Electric Power Systems Research 79 (2009) 210–219
[17] N. Bounar and all , “Adaptive fuzzy vector control
for a doubly-fed induction motor", Neuro-
computing151(2015)756–769
[18] O. ouledali , A .Meroufel , “Direct torque fuzzy
control of PMSM based on SVM”, Science Direct ,
Energy Procedia 74 ( 2015 ) 1314 – 1322
[19] Salloum G., Contribution at the Robust Control of
Doubly Fed Induction Motor, Thesis of National
Institue Polytechnic of Toulous, France 2007 .
[20] Naâmane Bounar, Abdesselem Boulkroune ,“
Adaptive Fuzzy Control of Doubly-Fed Induction
Machine” , Control Engineering And Applied
Informatics , Vol.16, No.2 pp. 98-110, 2014
[21] M. Abid, A. Aissaoui, H. Bounoua, “Sliding mode
application in speed and flux control of an induction
machine," Journal of Electrical and Electronics Engineering, vol. 6, no. 2, pp. 197– 204, 2006,
Istanbul university
[22] Rodic M, Jezernik K. Speed-sensorless sliding-mode
torque control of an induction motor. IEEE Trans Ind
Electron 2002;49:87–95.
[23] Yuan Li, Feng-You He, Zong-Bin Ye, Study on
Sliding Mode Speed Control with Fuzzy Approch for
Doubly-Fed Induction Motor, IEEE.Intrnational
Conference on Conference on Control and
Automation, Christchurch, New Zealand, December
9-11, p.2171-2175.
[24] Lekhchine, T. Bahi, “Speed control of doubly fed
induction motor" , Energy Procedia 74 ( 2015 ) 575 –
586
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 16 (2016) pp 9099-9105
© Research India Publications. http://www.ripublication.com
9105
[25] Piotr Pura, Grzegorz Iwanski, “Direct Power Control
with Referenced Torque and Reactive Power of
Doubly Fed Induction Machine Connected to
Unbalanced Grid ”, Ninth International Conference
on Ecological Vehicles and Renewable Energies
2014.
[26] G.Naveen, P.K.S.Sarvesh, B.Rama Krishna, “DTC
Control Strategy for Doubly Fed Induction Machine,
International Journal of Engineering and Advanced
Technology (IJEAT) ISSN: 2249 – 8958, Volume-
3(1), Oct 2013 .
[27] G. Abad, “Predictive direct control techniques of the
doubly fed induction machine for wind energy
generation applications”, PhD thesis, Mondragon
University, Arrasate, Spain, 2008.
[28] Zhang Y, Zhao Z, Zhu J, Xu W, Dorrell DG. Speed
sensorless direct torque control of 3-level inverter-
fed induction motor drive based on optimized
switching table. In: Conf. Rec. IEEE-IECON, Porto,
November; 2009. p. 1318-23.
[29] P. Werbos, “Beyond regression: new tools for
prediction and analysis in the behavioral sciences”,
PhD thesis, Cambridge, MA: Harvard University
Committee on Applied Mathematics, 1974.
[30] D. Kukolj, F. Kulic and E. Levi, “Design of speed
controller for sensorless electric drives based on ai
techniques: a comparative study”, Artificial
Intelligence in Engineering, vol. 14, 2000, pp. 165-
174.
[31] Luis A. Cabrera, Malik E. Elbuluk, and Donald
S.Zinger, “Learning Techniques to Train Neural
Networks as a State Selector for Inverter-Fed
Induction Machines Using Direct Torque Control”,
IEEE-PE, Vol. 12, No. 5, Sept. 1997, pp 788-799.
[32] X. Wu and L. Huang, “Direct torque control of three
level inverter using neural networks as switching
vector selector”, Industry Applications Conference.
Thirty-Sixth IAS Annual Meeting. Conference
Record of the IEEE, vol. 2, 2001, pp. 939-944.
[33] R. Kumar, R. A. Gupta, S.V. Bhangale, and
H. Gothwal, “Artificial neural network based
direct torque control of induction motor drives”,
IETECH Journal of Electrical Analysis, vol. 2, no. 3,
IETECH Publications 2008, pp.159-165.
[34] Cirrincione, M.C. Lu, and M. Pucci, “Direct torque
control of induction motors by use of the GMR neural
network”, International Joint Conference on Neural
Networks, vol. 3, 2003, pp. 2106-2111.