Proceedings of 5th International Symposium on

Intelligent Manufacturing Systems, May 29-31, 2006: 1170-1177

Sakarya University, Department of Industrial Engineering

FUZZY CONTROL OF A BUS SUSPENSION SYSTEM

Nurhan GÜRSEL1, İsmail Hakkı ALTAŞ2, Levent GÜMÜŞEL3

1. Karadeniz Technical University, Department of Mechanical Engineering, 61080

Trabzon, Turkey
Tel : 0 462 377 2907, Fax : 0 462 325 5526
e-mail: gnurhan@ktu.edu.tr

2. Karadeniz Technical University, Department of Electrical and Electronics

Engineering, 61080 , Trabzon, Turkey
Tel :0 462 377 2971, Fax : 0 462 377 7405
e-mail: ihaltas@altas.org

3. Karadeniz Technical University, Department of Mechanical Engineering, 61080

Trabzon, Turkey
Tel : 0 462 377 2944, Fax : 0 462 325 5526
e-mail: gumusel@ktu.edu.tr

Abstract— In this study fuzzy logic is used to control active suspension of a one-half-bus
model. The displacement is taken as input to the fuzzy logic controller and active forces,
improving ride comfort and handling properties are the controller outputs. The problem
was originally described and controlled using classical PID controllers in literature. A fuzzy
logic controller is designed and applied to this classical suspension problem in order to
minimize the oscillations and perform a smooth drive. The response of the system using
the proposed fuzzy logic controller is compared with those of PID controller given in
literature.

Index Terms — Dynamic modeling, Fuzzy control, Suspension system, State-

Space model
1. INTRODUCTION

Today, a struggling race is taking place among the automotive industry so as to

produce highly developed models. One of the performance requirements is advanced
suspension systems which prevent the road disturbances to affect the passenger comfort
while increasing riding capabilities and performing a smooth drive. While the purpose of
the suspension system is to provide a smooth ride in the car and to help maintain control
of the vehicle over rough terrain or in case of sudden stops, increasing ride comfort results
in larger suspension stroke and smaller damping in the wheel-hop mode [1]. Many control
methods have been proposed to overcome these suspension problems. Many active
suspension control approaches such as Linear Quadratic Gaussian (LQG) control,
adaptive control, and nonlinear control are developed and proposed so as to manage the
occurring problems [2-4]. Among the recent nonlinear control methods, fuzzy control
methods grab nowadays the attention of many researchers. A fuzzy model has excellent
capability in a nonlinear system description and is particularly suitable for the complex and
uncertain systems [5]. While there are various fuzzy modeling methods, Takagi– Sugeno
fuzzy models have been extensively studied up to now because the fuzzy model can
represent a nonlinear equation in spite of a small number of rules and be incorporated
with the conventional control methods easily.
During the last decades fuzzy logic has implemented very fast hence the first
paper in fuzzy set theory, which is now considered to be the seminal paper of the subject,
was written by Zadeh [6], who is considered the founding father of thefield. Then in 1975,
Mamdani, developed Zadeh`s work and demonstrated the viability of Fuzzy Logic Control
(FLC) for a small model steam engine. After this pioneer work, many consumer products
from washing machines to antibrake systems using fuzzy technology have been
developed and are currently available in the world. Støíbrskýl and the others [7], studied
on fuzzy logic control of active suspension of a one-half-car model. It is aimed to minimize
chassis and wheels deflection when road surfaces, pavements are acting on the tires of
running cars. At the end, the Matlab simulations are compared with a spring/damper
passive suspension. Huang and Lin [8], have studied on a quarter-car hydraulic
suspension system to evaluate the performance of active vehicle suspension, and a self
organizing fuzzy controller (SOFC) is used to control the position and acceleration
oscillation amplitudes of the spring mass due to the rough road variation.
In this study, an automatic suspension system for a ¼ bus is considered and a
fuzzy logic controller is designed when the bus is experiencing any road disturbance (i.e.
pot holes, cracks, and uneven pavement), the bus body should not have large oscillations,
and the oscillations should dissipate quickly. The road disturbance is simulated by a step
input and the distance between the body mass and simulation mass is the output of the
system.
The bus suspension problem used as a plant to be controlled by fuzzy logic here
was originally studied in the University of Michigan [9] and made public through the
internet. The original work is based on application of the classical PID controllers to
simulate the bus suspension system using Matlab functions. The state-space model of the
system given in [9] is combined with the proposed fuzzy logic controller here, and the
simulation of the bus suspension control system is carried one step ahead by improving
the controller’s effects.
The simulation results from the proposed fuzzy logic controller are compared with
those of given in [9] for both validation and shoving the improvement made.

1171
2.PROBLEM DESCRIPTION

2.1. State-Space Model

The one dimensional spring-mass-damper system given in Figure 1,

X1

Body mass

K1 b1
u

X2
Suspension
mass

K2
W b2

Figure 1. Model of ¼ bus suspension system

The followings are constants and variables of the system we are going to design :
• m1 = 2500 kg, body mass
• m2 = 320 kg, suspension mass
• k1=80,000N/m, spring constant of suspension system
• k2 = 500,000 N/m, spring constant of wheel and tire
• b1=350Ns/m. damping constant of suspension system
• b2=15,020 Ns/m. damping constant of wheel and tire
• u = force from the controller

To derive the dynamic equations of this system, Newton`s 2. law of motion is used, and
the equations below are presented.

M 1 X&&1 = − b1 ( X& 1 − X& 2 ) − K1 ( X 1 − X 2 ) + U

(1)
M 2 X&&2 = b1 ( X& 1 − X& 2 ) + K1 ( X 1 − X 2 ) + b2 (W& − X& 2 ) + K 2 (W − X 2 ) − U (2)

To transform the motion equations of the quarter-bus model into a state-space model, the
equation (3), including variable vector, input vector and the disturbance vector is formed
after some algebraic operations.
 X1 
& 
X U  Y = [C ][ X ] + [ D ]W
X& = [ A][ X ] + [ B ]W , Y = [ 0 0 1 0]  1  + [ 0 0]   ,
Y  W 
 1 
Y2 
 0 1 0 0  0 0 
 −b b b1 b1 b b K − b1   b 
 X1  
&  1 1 b2 
1 2
0 ( ( + 1 + 2 )− 1  X1 
 &&   M 1 M 2 M1 M1 M2 M2 M1 M1     M1 M 1M 2 
 X1     X& 1    U 
 &  =  b2 0 −(
b1 b b
+ 1 + 2 ) 1   
+  0 − 2
b  W 
 Y1   M 2 M1 M2 M2   Y1   M2  
 Y&     Y 2   
 2   K2 K1 K K  ( 1 + 1 ) − K 2 
M 0 −( + 1 + 2 ) 0   
 1 M1 M2 M2   M1 M2 M2 
(3)

1172
Since the distance X1-W is very difficult to measure, and the deformation of the tire (X2-
W) is negligible, we will use the distance X1-X2 instead of X1-W as the output in our
problem. The road disturbance (W) in this problem will be simulated by a step input. This
step could represent the bus coming out of a pothole.

2.2. Fuzzy Logic Controller

The main difference between classical control systems and fuzzy control is with
the classical mathematical methods it’s difficult to model and control the complex systems
but in fuzzy logic it’s easier and flexible, there are not precise boundaries.
The fuzzy logic comprised of three principal components: a fuzzification interface,
a knowledge base, and a defuzzification interface. The fuzzification interface transforms
crisp measured data into suitable fuzzy values. The knowledge base consists of a
database and a fuzzy control rule base computing the controller outputs in cope with the
rule base and data base. The defuzzification interface is responsible for converting these
fuzzy outputs into real-numbers.

Knowledge base

fuzzification Defuzzification

Decision Making

Nonfuzzy control input

Output and state
Plant

Figure 2.Typical architecture of fuzzy logic

The input and output membership functions of a fuzzy logic controller are usually
parametric functions such as triangular functions, trapezoidal functions, and bell-shaped
functions. A possible choice of the membership functions for the variables of the active
suspension system represented by a fuzzy set is as follows:

Figure 3. Triangular membership functions

The abbreviations used here correspond to:

• NL…..Negative Left • ZR…..Zero Right

• NM…..Negative Medium • PL……Positive Left
• NR……Negative Right • PM…..Positive Medium
• ZL……Zero Left • PR……Positive Right
• ZM…..Zero Medium

1173
The rule base used in the active suspension system for ¼ bus model consists of 9 rules
with fuzzy terms. The output of the controller is a fuzzy set of control. For defuzzification,
the method of ‘center of gravity’ is used here.

2.3. Fuzzy Control Scheme

The error signal, which is aimed to be minimized for a zero deflation level, is
obtained as the difference between desired deflation and the bounce of auto or deflation
from horizontal. The error and is one of the inputs to fuzzy controller. The other input is
the change in this error. Error signal is inputted to the controller to be minimized while the
change in error signal is used by fuzzy controller to get information about speed and
direction of the error in order to determine how fast and from which direction is the error
approaching toward zero. The proposed fuzzy controller uses this information to result in
the amount of increment; du. The process done by the fuzzy controller is depicted in
Figure 4.

Fig. 4. The block diagram of the fuzzy control system

The control input given to the system is u(k), which is obtained by adding current
increment du(k) to the previous value of the control input, u(k-1). This is nothing but
implementing fuzzy controller to get an output effect similar to that of an integrator without
any integrator present. The change du(k) in control input is equal to X1-X2, displacement
of the tire taken as the output of the system. By implementing the fuzzy decision rules an
appropriate output state is selected and assigned o a membership value. Then these truth
values are defuzzified to yield the final crisp change du(k). For each sampling the output
value will adjust the suspension system and then the control cycle will begin again to
generate the next value.

3. RESULTS

In this section the simulation results of the designed fuzzy logic controller are given. In
Figure 5, the step response of the ¼ bus model is given to display how the original open-
loop system performs without any feedback control.

1174
 -5
x 10 Step Response
2.5

1.5

Amplitude
1

0.5

0
0 10 20 30 40 50 60
Time (sec)

Figure 5. Step response of the open loop system

From the graph of the open-loop response for a unit step actuated force, we can see that
the system is under-damped. People sitting in the bus will feel very small amount of
oscillation and the steady-state error is about 0.013 mm. Moreover, the bus takes very
unacceptably long time for it to reach the steady state. The solution to this problem is to
add a controller into the system's block diagram to improve the performance. A classical
feedback controller used in [9] to damp the oscillations and reduce the settling time. The
closed loop response of the classical PID feedback control system with a 0.1 m
disturbance effect is given in Fig.6 where the maximum overshoot is approximately 0.0048
m and the settling time is about 1.5 seconds.

Fig 6. Results with classical PID controller

The results obtained using fuzzy logic controller are given in Figs 7 and 8. In Figure 7 the
fuzzy control response of the model with a 0.1 m step disturbance is given. The settling
time is about 1.0 second.

1175
 -4
x 10
6

y-(cm)
2

-2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-4
x 10 Time (s)
2

-2
e

-4

-6
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
4
x 10 Time (s)
2
u - Cont force

-1

-2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)
1
W - Distrbc(m)

0.5

-0.5

-1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Fig 7. Fuzzy control response

In order to get better views of both settling time and overshoot, the first graph in Fig. 7 is
enlarged and given in Fig. 8. As it can easily be seen from Fig. 8, the settling time has
been reduced from 1.5 s to about 1.0 s, while the overshoot is reduced from 0.0048 m
down to 0.0006 m.
-4
x 10
6

4
y-(cm)

-2
0 1 2 3 4 5
Time (s)

Fig. 8. System response with fuzzy controller

4. CONCLUSION

In this study, a fuzzy logic based controller is designed and employed for
controlling an active suspension system of a ¼ bus model. The proposed model is aimed
to developed and carry the response of classical PID controller up to a better level. By
using a nine-rule fuzzy rule base model with triangular fuzzy subsets, accepted
improvements are obtained in both settling time and oscillation reductions. A 25-rule fuzzy
controller could have given better results, however, increasing rule numbers is not
preferable since the number of the rules affect the processing time of the fuzzy algorithm.

1176
In a controller point of view any time delay affecting the operation of the system and
controller performance is not preferable. Therefore a nine-rule fuzzy controller is used due
to the acceptable results.

5. REFERENCES

1. H. Chen, Z. -Y. Liu, P.-Y. Sun , “Application of Constrained H_Control to Active

Suspension Systems on Half-Car Models”,Journal of Dynamic Systems,
Measurement, and Control, Vol. 127 / 353, SEP. 2005.
2. Gordon, T. J., Marsh, C., and Milsted, M. G., “A Comparison of Adaptive LQG
and Nonlinear Controllers for Vehicle Suspension Systems,” Veh. Syst. Dyn., 20,
1991,pp. 321–340.
3. Alleyne, A., and Hedrick, J. K.,“Nonlinear Adaptive Control of Active
Suspensions,” IEEE Trans. Control Syst. Technol., 3(1), 1995,pp. 94–101.
4. Ben Gaid, M., Cela, A.,Kocik, R., “Distributed control of a car suspension system,”
COSI - ESIEE - Cit´e Descartes,
5. Takagi, T. and Sugeno, M. Fuzzy identification of systems and its applications to
modeling and control, IEEE Trans. Systems Man Cybernet. 15(1) (1985), 116–132.
6. Zadeh, L. A., “Fuzzy sets, Information and Control 8 (1965), 338–353.
7. Støíbrskýl, A., Hyniová1, K., Honcù1, J. Kruczek1, A.,“Using fuzzy logic to
control active suspension system of one-half-car model ”, Acta Montanistica
Slovaca Ročník 8,2003, číslo 4.
8. Huang, S.,Lin, W.,“A Self-Organizing Fuzzy Controller for an Active Suspension
System”, Journal of Vibration and Control, Vol. 9, No. 9, 2003,1023-1040.
9. Control Tutorials for Matlab, The University of Michigan, the site as of January 23,
2006 is: http: //www.engin.umich.edu/group/ctm/

1177