SIMULATION AND EXPERIMENTAL ANALYSIS OF AN ACTIVE
VEHICLE SUSPENSION SYSTEM
SIMULATION AND EXPERIMENTAL ANALYSIS OF AN ACTIVE
VEHICLE SUSPENSION SYSTEM
A project report submitted in partial fulfilment of the
requirements for the award of the degree of
Master of Engineering (Mechanical)
v
ABSTRACT
This project was carried out to study the performance of a two degree-of-
freedom (DOF) active vehicle suspension system with active force control (AFC) as
the main proposed control technique. The overall control system essentially
comprises two feedback control loops. First is intermediate AFC control loop for the
compensation of the disturbances and second is the outermost Proportional-Integral-
Derivative (PID) control loop for the computation of the optimum commanded
force. Iterative learning method (ILM) and crude approximation (CA) were used as
methods to approximate the estimated mass in the AFC loop. Both simulation and
experimental studies were applied in this project. A quarter car model consists of
sprung and unsprung masses is considered in developing of the computer simulation
model in Simulink and also in the experimental set-up. Both simulation and
experimental work were carried out and the results between the two of them are
compared. The results of the simulation study show that active suspension system
using AFC with CA and ILM gives better performance compared to PID controller
and passive suspension system. Experimental results obtained in the study further
verified the potential and superiority of the performance of the active suspension
system with AFC strategy compared to the PID control.
vii
TABLE OF CONTENTS
CHAPTER TITLE PAGE
TITLE i
DECLARATION ii
DEDICATION iii
ACKNOWLEDGEMENT iv
ABSTRACT v
ABSTRAK vi
TABLE OF CONTENTS vii
LIST OF TABLES x
LIST OF FIGURES xi
LIST OF SYMBOLS xiv
LIST OF ABBREVIATIONS xvi
LIST OF APPENDICES xvii
1 INTRODUCTION 1
1.1 General Introduction 1
1.2 Objective 2
1.3 Scope of Work 2
1.4 Project Implementation 3
1.5 Organisation of Thesis 7
viii
2 THEORETICAL BACKGROUND AND
LITERATURE REVIEW
8
2.1 Introduction 8
2.2 Definition of Suspension System 8
2.3 Functions of a Vehicle Suspension 10
2.4 Types of Suspension System 11
2.4.1 Passive Suspension 12
2.4.2 Semi-active Suspension 13
2.4.3 Active Suspension 13
2.5 PID Controller 15
2.6 Active Force Control (AFC) 17
2.7 Iterative Learning Method 19
2.8 Review on Previous Research 20
2.9 Conclusion 21
3 MATHEMATICAL MODELLING AND
SIMULATION
23
3.1 Introduction 23
3.2 Quarter Car Model 23
3.3 Disturbance Models 26
3.4 Passive Suspension System Model 28
3.5 Active Suspension System Model 29
3.5.1 Active Suspension System with AFC-CA
Strategy
30
3.5.2 Active Suspension System Model with
AFC-ILM
31
3.6 Modelling and Simulation Parameters 33
3.7 Conclusion 37
4 SIMULATION RESULTS 35
4.1 Introduction 35
ix
4.2 Passive Suspension 35
4.3 Active Suspension 37
4.4 Active Suspension with AFC-CA 38
4.5 Active Suspension with AFC-ILM 40
4.6 Conclusion 41
5 EXPERIMENTAL SET-UP 42
5.1 Introduction 42
5.2 Simulink Model in Real-Time Workshop (RTW) 42
5.3 Experimental Set-up 47
5.3.1 Mechanical System 49
5.3.2 Electrical/Electronic Device 49
5.3.3 Computer Control 52
5.4 Parameters for Experiments 53
5.5 Conclusion 54
6 EXPERIMENTAL RESULTS AND DISCUSSION 55
6.1 Introduction 55
6.2 System Response Without Disturbance 56
6.3 System Response with the Sinusoidal Disturbance 59
6.4 System Response with the Step Disturbance 62
6.5 Conclusion 65
7 CONCLUSION AND RECOMMENDATION 66
7.1 Conclusion 66
7.2 Recommendation for Future Works 67
REFERENCES 68
APPENDICES 71
x
LIST OF TABLES
TABLE NO. TITLE PAGE
3.1 Parameters for suspension model 33
3.2 Simulation parameters 33
5.1 Suspension and pneumatic actuator parameters 53
xi
LIST OF FIGURES
FIGURE NO. TITLE PAGE
1.1 Flow chart of the project implementation 5
1.2 Gantt Chart of the project schedule 6
2.1 A suspension system 10
2.2 Passive suspension system 12
2.3 Semi-active suspension 13
2.4 Active suspension system 14
2.5 A block diagram of suspension system using PID controller 16
2.6 The schematic diagram of AFC strategy 18
2.7 A model of iterative learning method 19
3.1 Quarter car vehicle passive suspension 24
3.2 Quarter car vehicle active suspension 25
3.3 (a) Step input 27
3.3 (b) Bump and hole 27
3.3 (c) Sinusoidal 28
3.4 Simulink model of passive suspension system 29
3.5 Simulink model of active suspension system 30
3.6 Simulink model of active suspension system with AFC-CA 31
3.7 Simulink model of active suspension system with AFC-ILM 32
3.8 Subsystem of iterative learning method in AFC 32
4.1 (a) Passive suspension response to step input disturbance 36
4.1 (b) Passive suspension response to sinusoidal disturbance 36
4.2 (a) Active suspension response to step input disturbance. 37
4.2 (b) Active suspension response to sinusoidal disturbance 38
xii
4.3 (a) AFC-CA suspension response to step input disturbance 39
4.3 (b) AFC-CA suspension response to sinusoidal disturbance 39
4.4 (a) AFC-ILM suspension response to step input disturbance 40
4.4 (b) AFC-ILM suspension response to sinusoidal disturbance 41
5.1 Simulink model with RTW related to PID and AFC-ILM
control
43
5.2 Active suspension Simulink model in RTW 44
5.3 Pneumatic actuator subsystem 44
5.4 Body acceleration subsystem 45
5.5 Tyre acceleration subsystem 45
5.6 Disturbance subsystem 45
5.7 Suspension deflection system 46
5.8 Force tracking subsystem 46
5.9 AFC with ILM subsystem 47
5.10 Fotograph of the suspension system 48
5.11 The schematic of the experimental set-up 48
5.12 Accelerometer to measure body acceleration 50
5.13 Laser sensor to measure suspension deflection 50
5.14 LVDT to measure disturbance 51
5.15 Pressure sensor to measure actuator force 51
5.16 A computer set as the main controller 52
5.17 DAS 1602 interface card slotted in the CPU 53
6.1 Graph for body displacement response without disturbance 56
6.2 The close-up of body displacement response 57
6.3 Body displacement response without disturbance for B vary 58
6.4 Close-up body displacement response without disturbance
for B vary
58
6.5 Disturbance model type sinusoidal 59
6.6 Body displacement response with the sinusoidal disturbance 60
6.7 Body acceleration response with the sinusoidal disturbance 60
6.8 Suspension deflection response with the sinusoidal
disturbance
61
6.9 Tyre deflection response with the sinusoidal disturbance 61
6.10 Disturbance model type step. 62
xiii
6.11 Body displacement response with the step disturbance 63
6.12 Body acceleration response with the step disturbance 63
6.13 Suspension deflection response with the step disturbance 64
6.14 Tyre deflection response with the step disturbance 64
xiv
LIST OF SYMBOLS
a - Acceleration of the body
A - Proportional learning parameter
B - Derivative learning parameter
sb - Damping coefficient
D - Derivative
( )e t - Error (output – input)
( )e t� - Derivative error
af - Actuator force
aF - Actuated force
*F - Estimated force
I - Integral
dK - Derivative controller gain
iK - Integral controller gain
pK - Proportional controller gain
sk - Spring stiffness
tk - Tyre stiffness
( )m t - Control signal
sm - Sprung mass
um - Unsprung mass
*M - Estimated mass of the body
P - Proportional
kTE - Error value/current root of sum squared position track error
xv
ku - Current estimate value
1ku + - Next estimated value
rz - Displacement of road
sz - Displacement of sprung mass
uz - Displacement of unsprung mass
sz� - Velocity of sprung mass
uz� - Velocity of unsprung mass
sz�� - Acceleration of sprung mass
uz�� - Acceleration of unsprung mass
s uz z− - Deflection of suspension
u rz z− - Deflection of tyre
xvi
LIST OF ABBREVIATIONS
ADC - Analoque-to-digital converter
AF-AFC - Adaptive fuzzy active force control
AFC - Active force control
AFC-CA - Active force control with crude approximation
AFC-ILM - Active force control with iterative learning method
CA - Crude approximation
DAS - Data acquisition system
DCA - Digital-to-analoque converter
DOF - Degree of freedom
FLC - Fuzzy logic control
I/O - Input/output
IAFCRG - Intelligent Active Force Control Research Group
ILM - Iterative learning method
LVDT - Linear variable differential transformer
PC - Personal computer
PD - Proportional-Derivative
PID - Proportional-Integral-Derivative
PLC - programmable logic control
RTW - Real-Time Workshop
SANAFC - Skyhook and adaptive neuro active force control
xvii
LIST OF APPENDICES
APPENDIX TITLE PAGE
A Simulation Result for Various Conditions 71
B Experimental Results for Various Learning Parameter 72
C Experimental Results for Different Conditions 80
D The Sketch of the Experimental Rig 83
E The LVDT 84
F The Pressure Sensor 86
G The Data Acquisition System Card DAS 1602 88
H The Accelerometer 90
CHAPTER 1
INTRODUCTION
1.1 General Introduction
Traditionally, automotive suspension designs have been a compromise
between three conflicting criteria of road holding, load carrying and passenger
comfort. The suspension system must support the vehicle, provide directional
control during handling manouevres and provide effective isolation of passenger
payload from road disturbances [1]. Good ride comfort requires a soft suspension
wheras insentivity to applied load requires stiff suspension. Good handling requires
a suspension setting somewhere between the two.
Due to these conflicting demands, suspension design has had to be
something of a compromise, largely determined by the type of use for which the
vehicle was designed. Active suspensions are considered to be a way of increasing
the freedom one has to specify independently the characteristics of load carrying,
handling and ride quality.
2
A passive suspension system has the ability to storage energy via a spring
and to dissipate it via a damper. Its parameters are generally fixed, being chosen to
achive a certain level of compromise between road holding, load carrying and
comfort.
An active suspension system has the ability to store, dissipate and to
introduce energy to the system. It may vary its parameters depending upon operating
conditions and can have knowledge other than the strut deflection the passive system
is limited to.
1.2 Objective
The main objective of this project is to study the performance of an active
suspension system using active force control (AFC) through simulation and
experimental works.
1.3 Scope of work
The scope of this study consists of two major parts. The first is simulation
works and the second is experimental works. For the simulation works, the scope
involve is as follows:
i) To use an existing mathematical model of an active suspension.
ii) Apply active force control (AFC) with crude approximation (CA) and
3
iterative learning method (ILM) to active suspension system.
iii) Simulate active suspension system with active force control with
crude approximation (AFC-CA) and active force control with
iterative learning method (AFC-ILM) strategy incorporated with
different road profile.
iv) Study the performance of active suspension using AFC-CA and AFC-
ILM strategy compare to PID control.
The scopes involved in an experimental works is as follows:
i) Prepare experimental set-up.
ii) Develop Simulink model in Real-Time Workshop (RTW).
iii) Run experiment.
iv) Study the performance of active suspension system using AFC-ILM
strategy compare to PID control with the different disturbance.
v) Compare simulation results with the experimental results.
In experimental works, AFC strategy is used with iterative learning method
(ILM) is applied to approximate the estimated mass. The study in experimental
work will compare the result between PID and AFC-ILM only with different type
of disturbance. A quarter car model is considered in both simulation and
experimental study.
1.4 Project Implementation
The research is started with deriving the mathematical model of the main
dynamic system for the vehicle suspension system using Newton’s Second Law.
First, dynamic equation for passive suspension system is derived followed by active
4
suspension system. The model used is a two degree of freedom (DOF) system
representing a class of passenger car. Disturbances also were modeled
mathematically. Then, control scheme was developed and modelled. The schemes
include PID and AFC strategy employing both crude estimation and iterative
learning method.
Based on the derived models, a simulation study using MATLAB and
Simulink was carried out. Started with the passive suspension system with open loop
system followed by closed loop system of active suspension system. The results of
the simulations were then compared for both passive and active suspension. The
simulation results of active suspension system using AFC strategy and PID
controller also were compared.
Experimental set-up for the proposed system then was prepared. The work
involve during the set-up preparation is to develop experimental modules in the
MATLAB which known as Simulink model with Real-Time Workshop (RTW).
Then the experiment was carry out and the results obtained are analysed. Then
experimental results was compared to simulation results in order to validate the
results obtained for both method. This project implementation can be illustrated in a
form of a flow chart as shown in Figure 1.1. Gantt Chart of the project schedule is
shown in Figure 1.2.
5
Figure 1.1: Flow chart of the project implementation
6
W18
W18
W17
*
W17
*
W16
*
W16
*
W15
*
W15
*
W14
*
*
W14
*
Semester break
W13
*
W13
*
W12
*
*
W12
*
W11
*
*
W11
*
W10
*
*
W10
*
*
W9
*
*
W9
*
W8
W8
W7
*
*
W7
*
W6
*
*
W6
*
W5
*
*
W5
*
W4
*
W4
*
W3
*
W3
*
W2
*
W2
*
W1
*
W1
*
Activities
Brief idea
Literature review
Study dynamic system
Modelling proposed system
Simulate proposed system
Report writing
Presentation
Activities
Prepare experiments
Run experiments
Analyze result
Compare simulation results with
experimental results
Report writing
Presentation
SEMESTER 1
No.
1
2
3
4
5
6
7
SEMESTER 2
No.
1
2
3
4
5
6
Figure 1.2: Gantt chart of the project schedule
7
1.5 Organisation of Thesis
This thesis is organised into seven chapters. General introduction to the
suspension system, the objective and the scope of the project and how the project is
implemented is presented in Chapter 1. Chapter 2 discussed about theoretical information
and literature review related to the project backgraound. This includes the definition of
the suspension systems and its function. Explanation of types of suspension system and
the concept of proportional-integral-derivative (PID) controller, active force control
(AFC) and iterative learning method (ILM) also done in this chapter. A number of related
research is reviewed adequately in this chapter.
Mathematical modelling based on quarter car model is presented in Chapter 3.
Disturbances model and proposed simulation models also discussed in detail. Parameters
used for simulation is highlighted in this chapter. Chapter 4 presents the simulation
results for the different types of suspension system with various control strategies.
In chapter 5, experimental set-up for this project is explained in detail. This
includes the development of the Simulink model with Real-Time Workshop (RTW)
complete with all related subsystems in the model. Hardware components that used in the
set-up also described. Then, parameters for experiment is presented.
Chapter 6 presents experimental results. System response with various conditions
are presented and disscussed adequately. Chapter 7 gives the overall conclusion on the
study that has been done and recommend of future works could be considered as
extension to this study.
CHAPTER 2
THEORETICAL BACKGROUND AND LITERATURE REVIEW
2.1 Introduction
This chapter includes the study of the suspension definition, function of the
vehicle suspension and the vehicle dynamics. Three competing types of suspension
systems are also described in this chapter which are passive, semi active and active
suspension system. PID control, active force control (AFC) strategies and iterative
learning method (ILM) are also discussed. Then, the review of the previous
research related to the active suspension system is given.
2.2 Definition of Suspension System
Suspension system is a system that supports a load from above and isolates
the occupants of a vehicle from the road disturbances. Springs in the suspension
system are flexible elements. They able to store the energy applied to them in the
form of loads and deflections. They have the ability to absorb energy and bend when
9
they are compressed to shorter lengths. When a tyre meets an obstruction, it is forced
upward and the spring absorbs energy of this upward motion.
However, the spring absorbs this energy for a short time only and it will
release the energy by extending back to its original condition. When a spring
releases its stored energy, it does so with such quickness and momentum that the end
of the spring usually extends too far. The spring will go through a series of
oscillations, contractions and extension until all of the energy in the spring is
released. The natural frequency of the spring and suspension will determine the
speed of the oscillations.
The energy that released by spring is converted to heat and dissipated partly by
friction in the system by damper. Dampers usually in the form of piston working in
cylinders filled with hydraulic fluid. They exert a force which is proportional to the
square of the piston velocity. The function of damper is to restrain undesirable bounce
characteristic of the sprung mass. It also used to ensure the wheel assembly always
contact with the road by being excited at its natural vibration frequency.
Other mechanical elements in a suspension system are the wheel assemblies
and control geometry of their movement. Some of these elements are simple links
and multi-role members such as transverse torsion bars used to stabilize the vehicle
in corners by restricting roll. A suspension system comprises many elements that
include spring, damper, tyres, bushes, locating links and anti-roll bars are shown in
Figure 2.1.
10
Figure 2.1: A suspension system [2]
2.3 Functions of a Vehicle Suspension
A vehicle suspension system is a complicated system as it has to fulfill a large
number of partly contradictory requirements. Ride comfort, safety, handling, body
leveling and noise comfort are among the most important requirements that has to
fulfill.
Ride comfort can be determined by the acceleration of the vehicle body.
Acceleration forces are experienced by the passengers as a disturbance and set
demands on the load and the vehicle. The suspension system has the task to isolate
these disturbances from the vehicle body which caused by the uneven road profile.
The lower the acceleration, the better the rides comfort.
11
The safety of the vehicle during traveling is determined by the wheels ability
to transfer the longitudinal and lateral forces onto the road. The vehicle suspension
system is required to keep the wheels as close the road surface as possible. Wheel
vibration must be dampened and the dangerous lifting the wheels must be avoided. If
the dynamic forces occurring between the wheels and the road surface are small, the
braking, driving and lateral forces can be transferred to the road in an optimal
manner. The necessity of dampening the tyre system is the reason for the known
conflict of aims between comfortable and safety tuning.
Another function of the suspension system is the isolation of the vehicle body
from high frequency road disturbances. The passengers in the car note these
disturbances acoustically and thus the noise comfort is reduced. When there is
changes in loading, the suspension system has to keep the vehicle level as constant
as possible, so that the complete suspension travel is available for the wheel
movements. A lower suspension travel means that lower suspension working space
and this is a good suspension design. In order to fulfill all these contradict
requirements certain marginal conditions have to be considered.
2.4 Types of Suspension System
Generally there are three types of the suspension system. They are:
i) passive suspension
ii) semi-active suspension
iii) active suspension
12
2.4.1 Passive Suspension
Passive suspension system is the conventional suspension system. However
it is still to be found on majority of production car. It consists two elements namely
dampers and springs. The function of the dampers in this passive suspension is to
dissipate the energy and the springs is to store the energy. If a load exerted to the
spring, it will compress until the force produced by the compression is equal to the
load force. When the load is disturbed by an external force, it will oscillate around
its original position for a period of time. Dampers will absorb this oscillation so that
it would only bounce for a short period of time. Damping coefficient and spring
stiffness for this type of suspension system are fixed so that this is the major
weakness as parameters for ride comfort and good handling vary with different road
surfaces, vehicle speed and disturbances.
Figure 2.2: Passive suspension system
13
2.4.2 Semi-active Suspension
The element in the semi-active suspension system is same with passive
suspension system and it uses the same application of the active suspension system
where external energy is needed in the system. The difference is the damping
coefficient can be controlled. The fully active suspension is modified so that the
actuator is only capable of dissipating power rather than supplying it as well. The
actuator then becomes a continuously variable damper which is theoretically capable
of tracking force demand signal independently of instantaneous velocity across it
[3]. This suspension system exhibits high performance while having low system
cost, light system weight and low energy consumption.
Figure 2.3: Semi-active suspension
2.4.3 Active Suspension
The concept of active suspension system was introduced as early as 1958.
The difference compare to conventional suspension is active suspension system able
14
to inject energy into vehicle dynamic system via actuators rather than dissipate
energy. Active suspension can make use of more degrees of freedom in assigning
transfer functions and thus improve performance. The active suspension system
consists an extra element in the conventional suspension which is basically an
actuator that is controlled by a high frequency response servo valve and which
involves a force feedback loop. The demand foce signal, typically generated in a
microprocessor, is governed by a control law which is normally obtained by
application of various forms of optimal control theory [3]. Theoretically, this
suspension provides optimum ride and handling characteristics. It is done by
maintaining an approximately constant tire contact force, maintaining a level vehicle
geometry and by minimizing vertical accelerations to the vehicle. How ever due to
its complexity, cost and power requirements, it has not yet put into mass production.
Figure 2.4 shows an active suspension system.
An important issue in active suspension is energy consumption. It is
recognized that full active suspension, which must carry the full weight of vehicle,
would consume a considerable amount of energy and need high bandwidth actuators
(30 Hz) and control valves (100 Hz) [4,5]. Consequently, it was only installed in
some expensive and exclusive car or Formula One cars, and has not been mass
produced.
Figure 2.4: Active suspension system
15
2.5 PID Controller
PID control is a particular control structure that has become almost
universally used in industrial control. The letters ‘PID’ stand for Proportional,
Integral and Derivative. They have proven to be quite robust in the control of many
important applications for specific operating conditions. It structure is simple but
very effective feedback control method applied to dynamical systems. PID also most
conveniently integrated with other more advanced control techniques which more
often than not results in better overall performance. Pure PID control is excellent for
slow speed operation and with very small or no disturbances, the performance
severely degrades in the adverse conditions.
However, the simplicity of these controllers is also their weakness where it
limits the range of plants that they can control satisfactorily. Indeed, there exists a set
of unstable plants that they cannot be stabilized with any member of the PID family.
Nevertheless, the versatility of PID control ensures continued relevance and
popularity for this controller.
The PID method is error driven and largely relies on the proper tuning of the
controller gains and accurate information from the feedback element (sensor). The
basic algorithm of the PID is expressed as follows:
( ) ( ) ( ) ( )p i dm t K e t K e t dt K e t= + +∫ � (2.1)
where,
( )m t = control signal
pK = proportional controller gain
iK = integral controller gain
dK = derivative controller gain
16
( )e t = error (output – input)
( )e t� = derivative error
A simple PID controller applied to a vehicle suspension system can be
illustrated as shown in Figure 2.5.
Figure 2.5: A block diagram of suspension system using PID controller.
The effects of the P, I and D parameters to the system are as follows:
a) Proportional (P) action
This parameter provides a contribution which depends on the instantaneous
value of the control error. A proportional controller can control unstable plant but it
provides limited performance and non zero steady-state errors. This later limitation is
due to the fact that its frequency response is bounded for all frequencies.
b) Integral (I) action
Integral parameter gives a controller output that is proportional to the
accumulated error, which implies that it is a slow reaction mode. This
characteristic is also evident in its low-pass frequency response. The integral mode
plays a fundamental role in achieving perfect plant inversion at zero frequency.
17
This forces the steady-state error to zero in the presence of a step reference and
disturbance.
c) Derivative (D) action
Derivative action acts on the rate of change of the control error.
Consequently, it is a fast mode which ultimately disappears in the presence of
constant errors. It sometimes referred to as a predictive mode because of its
dependence on the error trend. The main limitation of the derivative mode is its
tendency to yield large control signals in response to high-frequency control errors,
such as errors induced by set-point changes or measurement noise.
2.6 Active Force Control (AFC)
Active force control strategy applied to dynamic system was proposed in the
early 80s by Hewit [6]. High robustness system can be achieved such that the system
remains stable and effective even in the presence of known or unknown
disturbances, uncertainties and varied operating conditions [7]. One of the succesful
AFC strategy applications is controlling robot arm, done by Mailah [8]. The study
has been demonstrated that AFC is superior compared to the conventional method in
controlling the robot arm.
The essence of the AFC is to determine the estimated force *F by measuring
two importants parameters. These parameters are the actuated force, aF (measured
by force sensor) and acceleration of the body, a (measured by accelerometer). An
appropriate estimation of the estimated mass of the body, *M was then multiplied
with the acceleration of the body, a yielding the estimated force. The mathematical
model for AFC can be written as follows:
18
* *aF F M a= − ⋅ (2.2)
If equation ( 2.2 ) can be fulfilled, it is expected that very robust system can
be achieved. Thus, it is the main aim of the study to apply the AFC method to
control a suspension effectively.
Figure 2.6 shows a schematic of the AFC strategy applied to a dynamic
system. Note that the estimated mass, *M in Figure 2.5 can be determined by a
number of methods such as crude approximation method, neural network, fuzzy
logic, iterative learning and genetic algorithms. However in this project, crude
approximation (CA) and iterative learning method (ILM) were used.
Figure 2.6: The schematic diagram of AFC strategy
19
2.7 Iterative Learning Method (ILM)
Iterative learning method (ILM) is one of the popular method in estimate the
next value. It has been applied to control a number of dynamic system [9]. As the
number of iteration increases, the track error converges to near zero datum and the
dynamic system is then said to operate effectively. In this project, the proposed
iterative learning algorithm takes the following form;
1 ( ) ( )k k k k
du u A TE B TE
dt+ = + + (2.3)
where
1ku + = next estimated value
ku = current estimate value
kTE = error value/current root of sum squared position track error
,A B = learning parameter
Figure 2.7 shows a graphical representation of the ILM algorithm.
A
1
Me(u k+1)
IM
IM(uk)
kdu/dt
Derivative
k
B
Add
1
TEk
Figure 2.7: A model of iterative learning method
20
2.8 Review on Previous Research
Active suspensions have been extensively studied nowadays compare to
passive suspension. Many researchers have studied and proposed a number of
control methods for vehicle suspensions. The first preview in the control of an active
system for a 1-DOF model was to introduced by Bender in 1967. Bender assumed
an integrated white noise terrain profile. He developed an optimal pair of damping
coefficient and spring stiffness by using Wiener filter theory to provide a wide range
of vibration isolation [10].
Tomizuka applied a discrete time, state space approach to Bender's problem
[11]. The optimal control scheme of that study involved both feedforward and
feedback elements. Tomizuka suggested his control logic could be realized in
practice by moving previewed samples through shift registers. The potential of the
preview control was demonstrated by subsequent studies for 2-DOF models by
Thomson.
In their paper, D’Amato and Viassolo described that the goal of this paper is
to minimize vertical car body acceleration, and to avoid hitting suspension limits
using fuzzy logic control (FLC) [12]. A controller consisting of two control loops is
proposed to attain this goal. The inner loop controls a nonlinear hydraulic actuator to
achieve tracking of a desired actuation force. The outer loop implements a FLC to
provide the desired actuation force. Controller parameters are computed by genetic
algorithm based optimization. The methodology proved effective when applied to a
quarter car model of suspension system.
Omar introduced a novel approach to control vehicle suspension system
using AFC strategy [13]. A proportional-derivative (PD) controller was incorporated
into the AFC control scheme. Crude approximation and iterative learning method
21
were used to estimated the initial mass in the AFC to effect the control action. The
simulation results have shown that the AFC is able to compensate the presence of
known or unknown disturbances to ensure that the system achieve the desired input.
Mailah and Priyandoko proposed an adaptive fuzzy active force control (AF-
AFC) to control vehicle active suspension system [14]. The technique proposed are
mainly for simplicity of the control low and to reduce the computational burden.
Non linear hydraulic actuator are used in the study. The simulation result shows the
performance of the proposed control method is found to be significantly superior
compared to the other systems considered in the study.
Priyandoko et al. introduced the practical design of a control technique apply
to a vehicle active suspension system [15]. Skyhook and adaptive neuro active force
control (SANAFC) are used as a control scheme. From the experimental result it
shows that SANAFC controller is very effective in isolating the vibration effects on
the sprung mass which in turn considerably improve the overall system
performance.
2.9 Conclusion
The theoretical backgrounds and previous research related to this study have
been outlined in this chapter. The information of the suspension in term of
definition, functions and type of suspension were adequately discussed. The
conventional PID controller and the fundamental concept of active force control
(AFC) applied to the dynamic system were also explained. Iterative learning method
(ILM) that will apply to estimate the estimated mass in AFC also discussed. It is
found that, many research papers discussed on optimization of various types of
22
suspension system to improve ride quality and road handling by using various types
of control schemes.
CHAPTER 3
MATHEMATICAL MODELLING AND SIMULATION
3.1 Introduction
In this chapter, a full modeling of the system dynamics related to the vehicle
suspension system, proposed control strategies and road disturbances are described.
This shall provide the basis for the rigorous computer simulation study to be carried
out using MATLAB and Simulink software package. The mathematical modelling
of the dynamic system is performed using the Newtonian mechanics. The suspension
system is modelled based on a quarter car configuration. The active suspension
system is specifically designed and modelled with the feedback control element
embedded into the system. A number of assumption that are made throughout the
modeling and simulation study is also described.
3.2 Quarter Car Model
Quarter car model are used to derive the mathematical model of the active
suspension system. The quarter car model is popularly used in suspension analysis
24
and design because it is simple to analyze but yet able to capture many important
characteristics of the full model. It is also realistic enough to validate the suspension
simulations.
Figure 3.1 shows a quarter car vehicle passive suspension system. Single
wheel and axle are connected to the quarter portion of the car body (sprung mass)
through a passive spring and damper. The tyre (unsprung mass) is assumed to have
only the spring feature and is in contact with the road terrain at the other end. The
road terrain serves as an external disturbance input to the system.
Figure 3.1: Quarter car vehicle passive suspension
The equations of motion for the the passive system are based on Newtonian
mechanics and given as: [14]
( ) ( )( ) ( ) ( )
s s s s u s s u
u u s s u s s u t u r
m z k z z b z z
m z k z z b z z k z z
=− − − −
= − + − − −
�� � �
�� � �
(3.1)
where
sm and um : sprung mass and unsprung mass respectively
sb : damping coefficient
25
sk and tk : stiffness of spring and tyre respectively
sz and uz : displacement of sprung mass and unsprung mass respectively
rz : displacement of road
s uz z− : deflection of suspension
u rz z− : deflection of tyre
sz� and uz� : velocity of sprung mass and unsprung mass respectively
sz�� and uz�� : acceleration of sprung mass and unsprung mass respectively
Active suspension system for a quarter car model can be constructed by
adding an actuator parallel to spring and dampe. Figure 3.2 shows a schematic of a
quarter car vehicle active suspension system.
Sprung MassMs
Ks
Unsprung MassMu
Kt
Bs
road profile
Zs
Zu
Zr
fa
Figure 3.2: Quarter car vehicle active suspension
The equations of motion for an active system are as follows:
( ) ( )( ) ( ) ( )
s s s s u s s u a
u u s s u s s u t u r a
m z k z z b z z f
m z k z z b z z k z z f
=− − − − +
= − + − − − −
�� � �
�� � �
Eq.(3.2)
where
af : actuator force
26
Some assumptions are made in the process of modeling the active suspension
system. The assumptions are:
i. the behaviour of the vehicle can be represented accurately by a quarter
car model.
ii. the suspension spring stiffness and tyre stiffness are linear in their
operation ranges and tyre does not leave the ground. The displacements
of both the body and tyre can be measured from the static equilibrium
point.
iii. The actuator is assumed to be linear with a constant gain.
3.3 Disturbance Models
There are three types of disturbances introduced to the vehicle suspension
system in this study. They are the step input, bump and hole, and sinusoidal
disturbance. Both bump and hole and sinusoidal are called the road disturbances
which represent the irregular road profile.
Figures 3.3 (a), (b) and (c) show the disturbances, step input, bump and hole
and sinusoidal respectively. The bump followed by a hole disturbance model is
adapted from the study by Roh and Park in [16] and the sinusoidal road input is
adapted from the work by Roukieh and Titli [17].
27
Figure 3.3 (a): Step input
Figure 3.3 (b): Bump and hole
28
Figure 3.3 (c): Sinusoidal
3.4 Passive Suspension System Model
The suspension system Simulink model is started basically with developing
the passive suspension system of a quarter car model. The dynamical system is
separated into two systems as the suspension system involves two degrees of
freedoms. This passive suspension model was modeled in Simulink form as shown
in Figure 3.4. This model was built based on the equation (3.1). There is an open
loop system with no feedback element for appropriate adjustment of parameters.
29
zrzu zszudotzuddot
zsdotzsddot
1/m2
mu
1/m1
ms
k2
ktk1
ks
b1
bs
XY GraphRoad profil e
1s
1s
1s
1s
du/dt
Derivative
Clock
Figure 3.4: Simulink model of passive suspension system
3.5 Active Suspension System Model
Active suspension system requires an actuator force to provide a better ride
and handling than the passive suspension system. The actuator force, Fa is an
additional input to the suspension system model. The model in Simulink was built
based on the equation (3.2) and shown in Figure 3.5. The actuator force is controlled
by the PID controller which involves a feedback loop.
30
zs
zu
zr
ref1/m2
mu
1/m1
ms
k2kt
k1
ks
b1
bs
-K-
actuator XY Graph
Road disturbance
PID
PID Controller
1s
1s
1s
1s
du/dt
Derivative
Clock
Figure 3.5: Simulink model of active suspension system
3.5.1 Active Suspension System Model with AFC-CA Strategy
Instead of using only PID controller, active suspension system in Simulink
model was further develop by introduced active force control with crude
approximation (AFC-CA) in the system. This model is shown in Figure 3.6. The
AFC-CA control Simulink blocks include the estimated mass gain, parameter 1/Ka
gain and the percentage of AFC application gain. The input to the AFC control is the
sprung mass acceleration and the output is summed with the PID controller output
before multiply with the actuator gain which finally results the generated actuator
force. Crude approximation method is used to estimated the estimated mass in the
AFC.
31
zs
zu
zr
ref1/m2
mu
1/m1
ms
k2kt
k1
ks
b1
bs
1
-K-
-K-
actuator XY Graph
Road disturbance
PID
PID Controller
-K-
Me
1s
1s
1s
1s
du/dt
Derivative
Clock
Figure 3.6: Simulink model of active suspension system with AFC-CA
3.5.2 Active Suspension System Model with AFC-ILM
To estimate the estimated mass for AFC, systematic method such as
intelligent method is appropriate to use rather than try and error. One of the
intelligent method is iterative learning method (ILM). This type of method applied
with AFC can be modelled as shown in Figure 3.7 .
32
zs
zu
zr
ref1/m2
mu
1/m1
ms
k2kt
k1
ks
b1
bs
1
-K-
-K-
actuator XY Graph
In1 Out1
Subsystem
Road disturbance
Product
PID
PID Control ler
1s
1s
1s
1s
y(n)=Cx(n)+Du(n)x(n+1)=Ax(n)+Bu(n)
Discrete State-Space1
du/dt
Derivative
Clock
Figure 3.7: Simulink model of active suspension system with AFC-ILM
Subsytem for iterative learning method is shown in Figure 3.8.
A
1
Me(u k+1)
IM
IM(uk)
kdu/dt
Derivative
k
B
Add
1
TEk
Figure 3.8: Subsystem of iterative learning method in AFC
33
3.6 Modelling and Simulation Parameters
The suspension parameters used in this study are adopted from the previous
study [15]. The detail of suspension model parameters are shown in Table 3.1 and
the simulation parameters are shown in Table 3.2.
Table 3.1: Parameters for suspension model.
Parameters Value
Sprung mass ( sm ) 170 kg
Unsprung mass ( um ) 25 kg
Spring stiffness ( sk ) 10,520 N/m
Damping coefficient ( sb ) 1,130 Ns/m
Tyre stiffness ( tk ) 86,240 N/m
Table 3.2: Simulation parameters
Parameters Value
Solver Ode45 (Dormand Prince)
Type Variable-step
Simulation time 10 s
Minimum step size Auto
Maximum step size Auto
Initial step size Auto
Relative tolerance 1e3
Absolute tolerance Auto
Zero crossing control use local setting
34
3.7 Conclusion
The mathematical equations of vehicle suspension system are derived using
quarter car model based on Newtonian mechanics. Then, the Simulink models for
passive and active suspension system were constructed. The disturbances also
modelled in the Simulink. The active suspension control systems in particular were
fully modelled complete with the control scheme with intelligent element to be
simulated to observe their responses. The simulation results for all models are
presented in the next chapter.
CHAPTER 4
SIMULATION RESULTS
4.1 Introduction
This chapter presents the simulated suspension responses results for all
suspension systems that are described in the previous chapter. The main concern of
the simulated suspension system responses results is the sprung mass displacement.
Comparisons of the results between types of the suspension system, different type of
disturbances and different type of control system are also discussed.
4.2 Passive Suspension
Figure 4.1 (a) and (b) show the response of passive suspension system to the
step and sinusoidal inputs respectively. Response shown for the step input is not
stable and need some time to settle down while under sinusoidal disturbance the
passive suspension could not adapt to the force given. This causes the sprung mass
displacement to occur for a long period of time.
36
0 1 2 3 4 5 6 7 8 9 100
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
Figure 4.1 (a): Passive suspension response to step input disturbance
0 1 2 3 4 5 6 7 8 9 10-1.5
-1
-0.5
0
0.5
1
1.5Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
Figure 4.1 (b): Passive suspension response to sinusoidal disturbance
37
4.3 Active Suspension
Figure 4.2 (a) and (b) show the response given by active suspension to the
step input and sinusoidal respectively. PID controller is used in this suspension and
it is a close loop system. PID is tuned optimisely so that the response for the step
input disturbance is good. However, for sinusoidal disturbance the response is not
good and not much different with the passive suspension. PID gain used are as
follows; Kp = 12, Ki = 5 and Kd = 4. These values are remain for the following sub
chapter 4.4 and 4.5 to observe their response.
0 1 2 3 4 5 6 7 8 9 100
0.2
0.4
0.6
0.8
1
1.2
1.4Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
Figure 4.2 (a): Active suspension response to step input disturbance
38
0 1 2 3 4 5 6 7 8 9 10-1.5
-1
-0.5
0
0.5
1
1.5Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
Figure 4.2 (b): Active suspension response to sinusoidal disturbance
4.4 Active Suspension with AFC-CA
Figure 4.3 (a) and (b) show the response given by active suspension with
AFC strategy and crude approximation method to the step input and sinusoidal
respectively. Both response, under step input and sinusoidal disturbance is much
better compare to PID controller only. AFC gives a good result although disturbance
is change. Estimated mass used in this simulation is 300 kg.
39
0 1 2 3 4 5 6 7 8 9 100
0.2
0.4
0.6
0.8
1
1.2
1.4Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
Figure 4.3 (a): AFC-CA suspension response to step input disturbance
0 1 2 3 4 5 6 7 8 9 100
0.2
0.4
0.6
0.8
1
1.2
1.4Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
Figure 4.3 (b): AFC-CA suspension response to sinusoidal disturbance
40
4.5 Active Suspension with AFC-ILM
Figure 4.4 (a) and (b) show the response of active suspension with AFC
strategy and iterative learning method to the step input and sinusoidal disturbance
respectively. The response given for both disturbance also as good as AFC-CA
suspension. This condition shows that the active suspension system with AFC
strategy still gives a good response although the disturbance is change. In other
words the active suspension with AFC is not affected by the changing of the
disturbance. Value of learning parameter A is set to 4 and B = 5. Initial condition
used is 200.
0 1 2 3 4 5 6 7 8 9 10-3
-2
-1
0
1
2
3
4
5
Time(s)
Dis
plac
emen
t (c
m)
Body Displacemnt
Figure 4.4 (a): AFC-ILM suspension response to step input disturbance
41
0 1 2 3 4 5 6 7 8 9 10-3
-2
-1
0
1
2
3
4
5
Time(s)
Dis
plac
emen
t (c
m)
Body Displacement
Figure 4.4 (b): AFC-ILM suspension response to sinusoidal disturbance
4.6 Conclusion
From the results it is proven that by using AFC, active suspension will
response much much better than without AFC. Passive suspension is the weakest
suspension to absorb any disturbance exerted to the system. Active suspension with
PID controller can give good performance if we can tune the PID controller gain
optimally. But when there is changes in the disturbances, PID controller is not
capable to compensate for that disturbance.
Active suspension with AFC strategy is proven not affected by the changing
of the disturbances. This means with AFC the high robust of suspension system can
be achieved. The system will remain stable and effective even in the presence of
known or unknown disturbance.
CHAPTER 5
EXPERIMENTAL SET-UP
5.1 Introduction
This chapter presents about the experimental work that was done in this
project. Experimental rig was developed using MATLAB, Simulink and Real-Time
Workshop after which a number of experiments were carried out.
This project used existing rig that was developed by the Intelligent Active
Force Control Research Group (IAFCRG). The quarter car rig was developed based
on the modified Perodua Kelisa suspension system. The details of the experimental
set-up will describe in this chapter.
5.2 Simulink Model in Real-Time Workshop (RTW)
Simulink model with Real-Time Workshop (RTW) developed in MATLAB
Software. This is the important element in the experiment because the RTW has an
43
ability to communicate with the outside world (suspension rig in this case) via an
interface card such as data acquisition system (DAS). DAS 1602 card is used in this
experiment as an interface card. With an aid of Simulink model in RTW and DAS
1602 card, the control of software and hardware of the rig is made possible through
the ‘hardware-in-the-loop’ concept.
Figure 5.1 shows the Simulink model with RTW that used in this
experiment. This model can be used for PID only control and also PID and AFC
with iterative learning method (AFC-ILM). A switch that is inserted into the middle
of the model will switch the system from the pure PID controller to the AFC-ILM
control scheme. Figure 5.2 shows the Simulink model with RTW for the physical
active suspension system.
tyre deflection
tyre acc
susp deflection
force
disturbance
body pos
body acc
time Active Suspension
PID
u-200
6.5
Iterative Learning-AFC
Figure 5.1: Simulink model with RTW related to PID and AFC-ILM control
44
7
body pos
6
tyre acc
5
force
4
susp deflc
3
disturbance
2
tyre deflection
1
body acceleration
ty re def lection
ty re acc
tyre accelerometer
Out1
susp def lc
susp deflection laser lvdt
to actuator desired press
pneumatic actuator
f orce
pressure1
force / pressure sensor
disturbance
Out2
disturbance lvdt
body position
body acceleration
body accelerometer
bddispl
bddispl.mat
uU( : )
|u|
Abs
-K-
AREA
1
to actuator
Figure 5.2: Active suspension Simulink model in RTW
Figures 5.3 to 5.8 show the subsystem models of the pneumatic actuator,
body acceleration, tyre accleration, disturbance model, suspension deflection and
force tracking model respectively.
1
desired press
desired pressure1
actuator
-K-
f(u)
f(u)
uf(u)
U( : )
Bad Link
butter
|u|1
to actuator
Figure 5.3: Pneumatic actuator subsystem
45
2
body acceleration
1
body position
bposbacc
bpos.matbacc.mat
1s
1s
uf(u) U( : )Bad Linkbutter
butterbutter
@1
Figure 5.4: Body acceleration subsystem
2
tyre acc
1
tyre deflection
acc
tdef
tacc
tdef.mat
tacc.mat
1s
1s
uf(u) U( : )Bad Linkbutter butterbutter
@3
Figure 5.5: Tyre acceleration subsystem
2
Out2
1
disturbance
distubance
distr
distr.mat
-1 f(u)f(u) U( : )Bad Linkbutter
@4
Figure 5.6: Disturbance subsystem
46
2
susp deflc
1
Out1
sus deflc
susdef
susdef.mat
-1
u
f(u)
U( : )
Bad Linkbutter
@7
Figure 5.7: Suspension deflection subsystem
2
pressure1
1
force
pressure
force1
press
force
press.mat
force.mat
1 f(u)
u
u
U( : )
U( : )
Bad Linkbutter
-K-
AREA
@8
Figure 5.8: Force tracking subsystem.
Figure 5.9 shows the most important element in the control system in the
experiment. That is active force control with iterative learning method (AFC-ILM)
simulink model.
47
2
pressure1
1
force
pressure
force1
press
force
press.mat
force.mat
1 f(u)
u
u
U( : )
U( : )
Bad Linkbutter
-K-
AREA
@8
Figure 5.9: AFC with ILM subsystem
5.3 Experimental Set-up
Figure 5.10 shows a photograph of an actual rig of active suspension system.
The schematic of the experimental set-up is shown in Figure 5.11.
Physical sensors required for input/output (I/O) signal were connected to a
PC-based data acquisition and control system using Matlab, Simulink and Real-
Time Workshop (RTW) that essentially constitute a hardware in the loop
configuration, implying that the simulation can be effectively converted to the
equivalent practical scheme without much fuss. A 100 Hz sampling frequency was
used in conjunction with a data acquisition card (DAS 1602) that is fitted into one of
the expansion slots of the personal computer (PC). Appropriate signals are processed
using the analoque-to-digital converter (ADC) and digital-to-analoque converter
(DCA) channels which are already embedded in the DAS card.
48
Figure 5.10: Photograph of the suspension system.
Figure 5.11: The schematic of the experimental set-up.
D/A
A/D
Suspension Test Rig PC-based control
MATLAB/CST/ Simulink/RTW
DAS1602 I/O card
to pneumatic actuator
from sensors (LVDTs, pressure sensor
& accelerometers)
PID, AFC and ILM,
Programmable Logic
Controller (PLC)
Disturbances
49
Accelerometers were installed at the sprung and unsprung mass of the
vehicle suspension system to measure body acceleration and tyre deflection. A laser
sensor was placed in the between of the sprung and unsprung mass to measure
suspension deflection. A linear variable differential transformer (LVDT) was used to
measure the vertical displacement of the road profile or disturbance. The
disturbances were used in this experiment was generated by a specially design
pneumatic system control by a programmable logic control (PLC).
The experimental set-up in this project is a mechatronics system. It is
because it involved an integration of the mechanical parts, electric/electronics
devices, and computer control to make the rig function.
5.3.1 Mechanical System
The mechanical system of the experimental set-up consists of the suspension
system itself as shown in Figure 5.10.
5.3.2 Electrical/Electronic Device
The electric/electronics devices were used in the experiment basically consist
of the sensors. Four types of sensors were used in the set-up, namely;
i. accelerometer
ii. laser sensors
iii. linear variable differential transformer (LVDT)
iv. pressure sensor
50
The location or position where all the sensors were placed can be seen in
Figures 5.12 to 5.15.
Figure 5.12: Accelerometer to measure body acceleration.
Figure 5.13: Laser sensor to measure suspension deflection
Accelerometer
Laser sensor
51
Figure 5.14: LVDT to measure disturbance.
Figure 5.15: Pressure sensor to measure actuator force.
LVDT
Accelerometer
Pressure sensor
52
All the signals from the sensors will be sent to the signal conditioners and
driver circuits. The circuits will process the signals to produce suitable signals to the
DAS Card.
5.3.3 Computer Control
The experimental set-up used a Pentium III computer as the main controller
with the software MATLAB/Simulink and RTW facility constituting the PC based
digital control. The DAS 1602 card is interfaced to the computer where the input and
output devices (actuators and sensors) were connected to the controller. Figure 5.16
shows the computer control system while Figure 5.17 shows the DAS 1602
interface card used in this experiment.
Figure 5.16: A Computer set as the main controller
53
Figure 5.17: DAS 1602 interface card slotted in the CPU.
5.4 Parameters for Experiments
Table 5.1 shows the suspension parameters and pneumatic actuator
parameters that have been used in experiment.
Table 5.1: Suspension and pneumatic actuator parameters.
Parameters Value
Sprung mass ( sm ) 170 kg
Unsprung mass (um ) 25 kg
Spring stiffness (sk ) 10520 N/m
Damping coefficient (sb ) 1130 Ns/m
Tyre stiffness ( tk ) 86240 N/m
Stroke length 116 mm
Diameter bore 40 mm
Ram area 0.0076 mm2
54
5.5 Conclusion
The Simulink model with RTW was successfully developed. The details of
the model are explained and all the subsystem models were clearly shown in this
chapter. Then, this Simulink model was integrated to the experimental rig constitutes
a full experimental set-up. A number of experiments were carried out. The results
obtained will be discussed in the next chapter.
CHAPTER 6
EXPERIMENTAL RESULTS AND DISCUSSION
6.1 Introduction
This chapter presents the results of the experiments that was carried out.
Same with the simulation part, in this experiment the main concerned of the
suspension system response result is the sprung mass or body displacement.
Comparisons of the results between different type of control scheme, there are PID
and AFC-ILM will be discussed in this section. The result for different type of
disturbances applied to the system also will presented. The results that are discussed
in this chapter were assume to give the best results obtained in the experiment using
the chosen parameters. Other results for different parameter setting were attached in
the appendix.
The results shown in this chapter are divided into two sections. For the first
200 second, the response belongs to PID controller. Then, for the next 200 seconds,
AFC-ILM control scheme take over. By doing this, we can see directly the different
responses (if any), displayed in single graph.
56
6.2 System Response Without Disturbance
Figure 6.1 shows the body displacement response of an active suspension
system without apply any disturbance into it. PID gain were used in this experiment
are, Kp = 35, Ki = 1.2 and Kd = 350. Learning parameter for the ILM are set as
follows; B=15 and Initial Condition = 25 kg. The value of learning parameter A is
set to vary. Results for the body acceleration, suspension deflection and tyre
deflection for the same conditions are shown in Appendix B.
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
A=10A=20A=50
PID
AFC-ILM
Figure 6.1: Graph for body displacement response without disturbance
Figure 6.2 shows the close-up of body displacement response of an active
suspension system without disturbance.
57
50 60 70 80 90 100-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
A=10A=20A=50
Figure 6.2: The close-up of body displacement response without disturbance
For this conditions, the results show that the learning parameter A=50 gives
the best results. However, there is no significant difference in result between pure
PID and AFC-ILM control schemes. The result looks almost similar. This means
that AFC-ILM control scheme was reached at the minimum level. Some tuning still
has to be done to get the better result for the AFC scheme.
Figure 6.3 shows the body displacement response with the fixed value of A
and varied value of B. The rest of the results for the variation of learning parameter
B can be found in Appendix B. Figure 6.4 shows the close-up of of the Figure 6.3.
58
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
B=15
B=20B=50
PID
AFC-ILM
Figure 6.3: Body displacement response without disturbance for B vary
40 50 60 70 80 90 100-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
B=15
B=20B=50
Figure 6.4: Close-up body displacement response without disturbance for B vary
59
6.3 System Response with the Sinusoidal Disturbance
Disturbances that applied to the active suspension system are step and
sinusoidal. The disturbance is generated by a specially design pneumatic system
controlled by PLC. Figure 6.5 shows the disturbance model and Figure 6.6 shows
the body displacement response to the sinusoidal disturbance. Sinusoidal signal that
gives to the system is high amplitude with high speed ( 2.8 Hz≈ ). Figures 6.7 - 6.9
show the response of the body acceleration, suspension deflection and tyre
deflection respectively to the sinusoidal disturbance.
50 55 60 65 70 75 80 85 90 95 1001.5
1.6
1.7
1.8
1.9
2
2.1
2.2
2.3Sinusoidal Disturbance
Time(s)
Am
plitu
de (
cm)
Figure 6.5: Disturbance model type sinusoidal
60
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
PID
AFC-ILM
Figure 6.6: Body displacement response with the sinusoidal disturbance
0 50 100 150 200 250 300 350 400-15
-10
-5
0
5Body Acceleration
Time(s)
Acc
eler
atio
n (m
/s2 )
PID
AFC-ILM
Figure 6.7: Body acceleration response with the sinusoidal disturbance
61
0 50 100 150 200 250 300 350 400-2
-1
0
1
2
3
4
5
6
7
8Suspension Deflection
Time(s)
Def
lect
ion
(cm
)
PID
AFC-ILM
Figure 6.8: Suspension deflection response with the sinusoidal disturbance
0 50 100 150 200 250 300 350 400-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Tyre Deflection
Time(s)
Def
lect
ion
(cm
)
PID
AFC-ILM
Figure 6.9: Tyre deflection response with the sinusoidal disturbance
62
6.4 System Response with the Step Disturbance
The disturbance type step also generated by a specially design pneumatic
system controlled by PLC. Figure 6.10 shows the disturbance model of the step.
The shape of step is not so good due to some leaking at the pneumatic system. It
caused the pneumatic system cannot hold the load for a period of time to form a
good step. However the disturbance produce still can be used as long as we can put
some interruption to the system and observe the response. All the results observed
are shown in the Figures 6.11-6.14.
10 20 30 40 50 60 700
1
2
3
4
5
6
7
8Step Disturbance
Time(s)
Ste
p (c
m)
Figure 6.10: Disturbance model type step
63
0 50 100 150 200 250 300 350 400-4
-3
-2
-1
0
1
2
3
4Body displacement
Time(s)
Dis
plac
emnt
(cm
)
PID
AFC-ILM
Figure 6.11: Body displacement response with the step disturbance
0 50 100 150 200 250 300 350 400-30
-25
-20
-15
-10
-5
0
5Body Acceleration
Time(s)
Acc
eler
atio
n (m
/s2 )
PID
AFC-ILM
Figure 6.12: Body acceleration response with the step disturbance
64
0 50 100 150 200 250 300 350 4000
1
2
3
4
5
6
7
8Suspension Deflection
Time(s)
Def
lect
ion
(cm
)
PID
AFC-ILM
Figure 6.13: Suspension deflection response with the step disturbance
0 50 100 150 200 250 300 350 400-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Tyre Deflection
Time(s)
Def
lect
ion
(cm
)
PID
AFC-ILM
Figure 6.14: Tyre deflection response with the step disturbance
65
The results for other conditions, i.e different value of parameter A and B,
different values of Initial Condition and the different type of disturbance, please
refer to Appendix C.
6.5 Conclusion
In order to get the best tune for the learning parameter, A and B, the
experiments were carried out without apply any disturbance to the suspension
system. The system with the set learning parameter then was applied the
disturbances. The result show that the active suspension system with pure PID
controller gives almost similar response with the AFC-ILM control scheme. AFC-
ILM suppose to give better response than pure PID. It means that AFC-ILM control
scheme was reached at the minimum level. Some tuning still has to be done to get
the better result for the AFC scheme but due to time constraint existing learning
parameters are remained for this project.
CHAPTER 7
CONCLUSION AND RECOMMENDATION
7.1 Conclusion
The implementation of the active force control (AFC) to the vehicle
suspension system has been successfully done in simulation study and in
experimental work. In simulation study the result shows that the use of AFC make
the system robust. It is because AFC can compensate any internal and external
disturbances that presence in the suspension system. In experimental work, it should
show the same result. However due to highly skill needed to tune the learning
parameter in the ILM to estimate initial mass for AFC, the result obtained is just
same with the pure PID controller.
The study in simulation demonstrate that AFC-CA and AFC-ILM give better
performance compare to pure PID controller. The most important thing in AFC is to
estimate the initial mass. If we get the accurate approximate initial mass, AFC will
give a better performance. In estimating the initial mass the method use is crude
approximation and iterative learning. Crude approximation method is easier than
iterative learning as we just have to directly change the value of initial mass. This
method however will take long time to get the right value of initial mass. Iterative
learning method is more intelligent to estimate the initial mass value as it will iterate
67
repetitively by decrease the error until it get the right value. But the problem in this
method is to tune the learning parameter.
Simulation study shows that by using AFC control scheme, the performance
of the system (active suspension in this case) will improve tremendously. AFC able
to compensate the presence of the known or unknown disturbances.
In experimental works, the experiment was run using the active suspension
rig. Two type of control schemes were used, those are PID and AFC-ILM and the
results from both were compared. From the experimental work, the results show that
the performance of the suspension system for both control scheme almost similar.
The theory says that AFC is better than PID. This condition was happen due to
learning parameter tuning for ILM is still not satisfy. Fine tune the learning
parameter to the right value will change to the better result.
7.2 Recommendation for Future Works
There are few number of future works could be considered as an extension to
the present study. They are as follows;
i) consider the use of the percentage AFC to the system.
ii) the use of self tuning method to estimate the estimated mass in AFC.
iii) study the effects to the performance of suspension system by increase
the sprung mass load.
68
REFERENCES
[1] P.G. Wright. (1984). The Application of Active Suspension to High
Performance Road Vehicles, Microprocessors in Fluid Engineering
IMechE Conference Publications.
[2] http://www.lanciamontecarlo.net/Scorpion/Technical_Suspension.html
[3] D. A. Crolla. (1988). Theoretical Comparisons of Various Active
Suspension Systems in Terms of Performance and Power Requirements. in
“Advanced Suspensions”, Suffolk : Mechanical Engineering Publications
Limited. pp 1 – 9.
[4] Alleyne, A., Neuhaus, P.D., Hedrick, J.K (1993). Application of Non
Linear Control Theory to Electronically Controlled Suspension, Vehicle
System Dynamics Vol. 22, No. 5-6, P.309-320.
[5] Gopalasamy, S,. et. al.. (1997). Model Predictive Control For Active
Suspension. Controller Design and Experimentally Study. Trans. of ASME,
Journal of Dynamic Systems and Control, Vol. 61, pp. 725-733.
[6] Hewit, J.R., (1998). Advances in Teleoperations, Lecture note on Control
Aspects, CISM.
[7] Mailah, M. and Yong, M.O., (2001). Intelligent Adaptive Active Force
Control of a Robot Arm With Embedded Iterative Learning Algorithms,
Jurnal Teknologi, UTM, No.35(A), pp. 85-98.
69
[ 8] Musa Mailah. (1999). A Simulation Study on the Intelligent Active Force
Control of A Robot Arm Using Neural Network, Jurnal Teknologi (D),
Universiti Teknologi Malaysia. pp 55 – 78.
[9] Arimoto, S., Kawamura, S., and Miyazaki, F. (1986). Convergence,
Stability and Robustness of Learning Control Schemes for Robot
Manipulators, Recent Trends in Robotics: Modelling, Control and
Education, ed. by Jamshidi M., Luh L.Y.S., and Shahinpoor M. 307 – 316.
[10] Zhang, Y. (2003). A hybrid adaptive and robust control methodology with
application to active vibration isolation, University of Illinois, Urbana-
Champaign, Ph.D. Thesis.
[11] Baillie, A.S. (1999). Development of a fuzzy logic controller for an active
suspension of an off-road vehicle fitted with terrain preview, Royal Military
Collage of Canada, Kingstone, Canada, Ph.D. Thesis.
[12] D’Amato, F. J. and Viassolo, D. E. (2000). Fuzzy Control for Active
Suspensions, Mechatronics, 10: 897-920.
[13] Omar, Z. (2002). Modelling and Simulation of an Active Suspension System
Using Active Force Control Strategy, MSc. Project Report, Universiti
Teknologi Malaysia.
[14] Mailah M., Priyandoko G. (2005). Simulation of a Suspension System with
Adaptive Fuzzy Active Force Control, International Journal of Simulation
Modelling, Vol.6 No.1, pp 25-36.
[15] G. Priyandoko, et al. (2007). Skyhook Adaptive Neuro Active Force Control
for an Active Suspension System, Procs. Of CIM07, Johor Persada
Convention Centre.
70
[16] Hyoun-Surk Roh and Youngjin Park. (1999). Preview Control of Active
Vehicle Suspension Based on a State and Input Estimator, in Ronald K.
Jurgen (Ed.).“Electronic Steering and Suspension Systems.” Warrendale :
Society of Automotive Engineers, Inc. pp 277 – 284.
[17] S. Roukieh and A. Titli. (1992). On the Model-Based Design of Semi-Active
and Active Suspension for Private Cars, in “Total Vehicle Dynamics.”
London : Mechanical Engineering Publications Limited. pp 305 – 318.
71
APPENDIX A
Simulation Result for Various Conditions
a) Passive suspension response for different value of step input
0 1 2 3 4 5 6 7 8 9 100
2
4
6
8
10
12
14
16
18
time (s)
step
inpu
t va
lue
(cm
)
Response for different step input for passive suspension
1.0
2.010.0
b) Active suspension response for different value of proportional gain (kp)
0 2 4 6 8 10-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3
time, t
Am
plitu
de
zs response with PID controller (various kp, ki=5,kd=4)
kp=12
kp=20
kp=30
72
APPENDIX B
Experimental Results for Various Learning Parameter
The graphs show the response and their close-up for body displacement,
body acceleration, suspension deflection and tyre deflection respectively.
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
A=10
A=20A=50
50 60 70 80 90 100-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
A=10
A=20A=50
73
0 50 100 150 200 250 300 350 400-30
-25
-20
-15
-10
-5
0
5
10Body Acceleration
Time(s)
Acc
eler
atio
n (m
/s2 )
A=10
A=20A=50
0 10 20 30 40 50 60 70 80 90 1000
1
2
3
4
5
6Body Acceleration
Time(s)
Acc
eler
atio
n (m
/s2 )
A=10
A=20A=50
74
0 50 100 150 200 250 300 350 400-6
-4
-2
0
2
4
6
8Suspension Deflection
Time(s)
Def
lect
ion
(cm
)
A=10
A=20A=50
0 10 20 30 40 50 60 70 80 90 1000
1
2
3
4
5
6
7Suspension Deflection
Time(s)
Def
lect
ion
(cm
)
A=10
A=20A=50
75
0 50 100 150 200 250 300 350 400-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Tyre Deflection
Time(s)
Def
lect
ion
(cm
)
A=10
A=20A=50
0 10 20 30 40 50 60 70 80 90 100-1
-0.5
0
0.5
1
1.5Tyre Deflection
Time(s)
Def
lect
ion
(cm
)
A=10
A=20A=50
76
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
B=15
B=20B=50
40 50 60 70 80 90 100-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Body Displacement
Time(s)
Dis
plac
emen
t (c
m)
B=15
B=20B=50
77
0 50 100 150 200 250 300 350 400-30
-25
-20
-15
-10
-5
0
5
10Body Acceleration
Time(s)
Acc
eler
atio
n (m
/s2 )
B=15
B=20B=50
0 10 20 30 40 50 60 70 80-5
-4
-3
-2
-1
0
1
2
3
4
5
6Body Acceleration
Time(s)
Acc
eler
atio
n (m
/s2 )
B=15
B=20B=50
78
0 50 100 150 200 250 300 350 400-6
-4
-2
0
2
4
6
8Suspension Deflection
Time(s)
Def
lect
ion
(cm
)
B=15
B=20B=50
10 20 30 40 50 60 70 80 90 100 110 120-2
-1
0
1
2
3
4
5
6
7Suspension Deflection
Time(s)
Def
lect
ion
(cm
)
B=15
B=20B=50
79
0 50 100 150 200 250 300 350 400-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Tyre Deflection
Time(s)
Def
lect
ion
(cm
)
B=15
B=20B=50
0 10 20 30 40 50 60 70 80 90 100-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1Tyre Deflection
Time(s)
Def
lect
ion
(cm
)
B=15
B=20B=50
80
APPENDIX C
Experimental Results for Different Conditions
Data 1 : A=100, B=90, IC=250
Data 2 : A=150, B=200, IC=500
Disturbance type : High Sin and high speed ( ≈1.8Hz)
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Body displacement
Time(s)
Dis
plac
emnt
(cm
)
Data1
Data2
PID
AFC-ILM
81
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Body Acceleration
Time(s)
Acc
eler
atio
n (m
/s2 )
Data1
Data2
PID
AFC-ILM
0 50 100 150 200 250 300 350 4000
1
2
3
4
5
6
7
8Suspension Deflection
Time(s)
Def
lect
ion
(cm
)
Data1
Data2
PID
AFC-ILM
82
0 50 100 150 200 250 300 350 400-5
-4
-3
-2
-1
0
1
2
3
4
5Tyre Deflection
Time(s)
Def
lect
ion
(cm
)
Data1
Data2
PID
AFC-ILM
83
APPENDIX D
The Sketch of the Experimental Rig
Load
Tyre
Body
Pneumatic to generate
disturbanceMotor
Pneumatic actuator
84
APPENDIX E
The LVDT
The LVDT used in this project is AML/IEU+/-75mm-X-10.
85
86
APPENDIX F
The Pressure Sensor
The pressure sensor used in this project is model DP2-22.
87
88
APPENDIX G
The Data Acquisition System Card DAS 1602
89
90
APPENDIX H
The Accelerometer
The accelerometer used in this project is ADXL-105EM-1.
91
92
93
94
95
96
97