Sie sind auf Seite 1von 6

Modeling and Control Active Suspension System for

a Full Car Model


Rosheila Darus
1
, Yahaya Md. Sam
2
1
Faculty of Electrical Engineering, Universiti Teknologi Mara (UiTM),
Cawangan Terengganu, 23000 Dungun, Terengganu, MALAYSIA
2
Faculty of Electrical Engineering, Universiti Teknologi Malaysia (UTM),
81310 Skudai, Johor, MALAYSIA
Abstract-The purpose of this paper is to investigate the
performance of a full car model active suspension system using
LQR controller. Dynamic model used in this study is a linear
model. A linear model can capture basic performances of vehicle
suspension such as body displacement, body acceleration, wheel
displacement, wheel deflection, suspension travels, pitch and
yawn. Performance of suspension system is determined by the
ride comfort and vehicle handling. It can be measured by car
body displacement and wheel displacement performance. Two
types of road profiles are used as input for the system.
Simulation is based on the mathematical model by using
MATLAB/SIMULINK software. Results show that the
performance of body displacement and wheel displacement can
be improved by using Linear Quadratic Regulator control
(LQR).
I. INTRODUCTION
Suspension is a common property to all automobiles. Its
isolated the vehicle body from road disturbances for
comfortable ride. Conventional suspension system consists of
a spring and damper, also known as passive suspension. In
active suspension system control strategy is a very important.
With correct control strategy, it will give better compromise
between ride comfort and vehicle handling. Nowadays there a
lot of researches have been done to get more efficient and
better performance of active suspension by having suitable
control strategy. In most of the research has been done, a
linearized model is used. One of the reason is it can be
derived easily and it can capture basic features of a real
vehicle problem. Most of the studies are concerning on
controlling the forces created by the suspension damper and
springs.
Application of the LQR method to the active suspension
system has been proposed [1]. In this study, the LQR method
is used to improve the vehicle handling and the ride comfort
for a quarter car model.
II. DYNAMIC MODELING
Many ongoing research studies about designing a control
system have used the well know 7-DOF as shown in Fig. [1].
Mathematical modeling of the system obtained from [2].
Tf Tr
roll axis
pitch axis
a
b
mu1
mu4
mu2
mu3
zr1
zr2
zr4
zr3
Zs
ms
Zu2
Zu3
Zu4
Zu2
Fig. 1. Suspension System for Full Car Model
The dynamic model derived by [2] divided into 7 main parts:
Bouncing of the sprung mass
1 1 2 2 3 3
4 4 1 1 2 2
3 3 4 4 1 2 3 4
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
s s s u s u s u f
s u s u s u
s u s u
m Z b Z Z bf Z Z br Z Z
br Z Z kf Z Z kf Z Z
kr Z Z kr Z Z u u u u
=

+ + + +


(1)
Pitching of the sprung mass
1 1 2 2 3 3
4 4 1 1 2 2
3 3 4 4 1 2 3 4
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
p s r s u s u s u f f
r s u s u s u f f
r r s u s u
I b a Z Z b a Z Z b b Z Z
b b Z Z k a Z Z k a Z Z
k b Z Z k b Z Z au au bu bu
=
+ +
+ + + +


(2) (2)
Rolling of the sprung mass
1 1 2 2 3 3
4 4 1 1 2 2
3 3 4 4 1 2 3 4
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
r s r r s u s u s u f f f f
r r s u s u s u f f f f
r r r r r r s u s u f f
I b T Z Z b T Z Z b T Z Z
b T Z Z k T Z Z k T Z Z
k T Z Z k T Z Z T u T u T u T u
=
+ +
+ + +


(3)
Vertical Direction for each wheel
1 1 1 1 1 1 1 1
( ) ( )
u s u s u u r uf f f tf tf
m Z b Z Z k Z Z k Z u k Z = + +
(4)
2 2 2 2 2 2 2 2
( ) ( )
u s u s u u r uf f f tf tf
m Z b Z Z k Z Z k Z u k Z = + +
(5)
3 3 3 3 3 3 3 3
( ) ( )
ur r r tr tr u s u s u u r
m Z b Z Z k Z Z k Z u k Z = + +
(6)
4 4 4 4 4 4 4 4
( ) ( )
ur r r tr tr u s u s u u r
m Z b Z Z k Z Z k Z u k Z = + +
(7)
13
2009 5th International Colloquium on Signal Processing & Its Applications (CSPA)
978-1-4244-4152-5/09/$25.00 2009 IEEE
where;
1 s s s s f
Z T a Z = + +
1 s s s s f
Z T a Z = + +

2 s s s s f
Z T a Z = + +
2 s s s s f
Z T a Z = + +

3 r s s s s
Z T b Z = +
3 r s s s s
Z T b Z = +

4 r s s s s
Z T b Z = +
4 r s s s s
Z T b Z = +

Fig. 1 is an accurate 7-DOF where linearization and model
reduction technique are used to find a simple version of a
complex full car model. There are roll, pitch and the vertical
displacement of sprung mass and four unsprung mass are
included.
TABLE 1
DEFINITIONS OF VARIABLES
Variables Definitions
s
Z
Vertical displacement
s
Z
Vertical velocity
1 u
Z
Vertical displacement of front right wheel
1 u
Z
Vertical velocity of front right wheel
2 u
Z
Vertical displacement of front left wheel
2 u
Z
Vertical velocity of front left wheel
3 u
Z
Vertical displacement of rear right wheel
3 u
Z
Vertical velocity of rear right wheel
4 u
Z
Vertical displacement of rear left wheel
4 u
Z
Vertical velocity of rear left wheel
s

Pitch angle
s

Pitch rate
s

Roll angle
s

Roll rate
III. CONTROLLER DESIGN
Assume quadratic performance index in the form of
1
( )
2
0
t
T T
J x Q x u R u d t = +
}
(8)
Matrix Q is symmetric positive semi-definite and R is positive
symmetric definite.
Matrix K is presented as;
1
' K R B P

= (9)
Linear feedback control law is obtained as;
( ) ( ) u t Kx t =
(10)
Designed matrix K is determined by using MATLAB by
setting suitable matrix Q and R.
IV. SIMULATION
Input disturbance or irregular excitation from the road
surface divided into 2 types; single bump and two bumps. It
can show the ability of LQR to adapt different types of road
disturbance.
Case 1: A single bump input shows in (11) for front wheel and
(12) for rear wheel. Assume that front right and left wheel
reached bump at the same time.
( )
(1 cos 8 ), 0.5 0.75
0 otherwise
d t
a t s t s

=
s s (11)
( )
(1 cos 8 ), 3.0 3.25
0 otherwise
d t
a t s t s

=
s s (12)
Fig. 1. Input Disturbance of a Single Bump for Front Right and Left Wheel.
Fig. 2. Input Disturbance of a Single Bump for Rear Right and Left
Case 2: Two bumps input shows in (13) for front wheel and
(14) for rear wheel. Assume that front right and left wheel
reached bump at the same time
14
( )
(1 cos 8 ), 0.5 0.75
(1 cos 8 ) /2 6.5 6.75
0 otherwise
d t
a t s t s
a t s t s

=
s s
s s

(13)
( )
(1 cos 8 ), 3.0 3.25
(1 cos 8 ) /2 9.0 9.255
0 otherwise
d t
a t s t s
a t s t s

=
s s
s s

(14)
Fig. 3. Input Disturbance of a Two Bumps for Front Right and Left

Fig. 4. Input Disturbance of a Two Bumps for Rear Right and Left
Parameters for the full car model of active suspension
systems shows in table 1 [3].
TABLE 2
CAR PARAMETER
mass of the car body or sprung mass, m
s
1500 kg
front mass of the wheel or unsprung
mass, m
uf
59 kg
rear mass of the wheel or unsprung mass,
m
ur
59 kg
Pitch of moment of inertia, I
p
2160 kg m
2
roll of moment of inertia, I
r
460 kg m
2
stiffness of car body spring for front, k
f
35000 N/m
stiffness of car body spring for rear, k
r
38000 N/m
front treat, T
f
0.505 m
rear treat, T
r
0.557m
font and rear tire stiffness, k
tf
and
ktr
190000 N/m
front damping , b
f
1000 N/m
rear damping, b
r
1100 N/m
c.g to front wheel, a 1.4m
c.g to rear wheel, b 1.7 m
V. RESULTS
For comparison purpose, the performance of active
suspension system using LQR technique is compared with
passive suspension system.
Fig. 5. Force Generate for Front Right and Left Actuator by Using LQR
controller (Case 1)
15
Fig. 6. Force Generate for Rear Right and Left Actuator by Using LQR
controller (case 1)
Fig. 5 and Fig. 6 show force generated by actuators for
each suspension sets for road profile with single bump. Front
right and front left give same amount of force due to input
disturbance is the same. Rear right and rear left also have
same types of disturbance; therefore force generated at both
actuators for front wheels are same and same condition happen
to rear actuators.
Fig. 7. Body Displacement (case 1)
Fig. 8. Wheel Displacement for Front Right and Left (case 1)
Fig. 9. Wheel Displacement for Rear Right and Left (case 1)
Fig. 10. Suspension Travel Front Right and Left (case 1)
Fig. 11. Suspension Travel Rear Right and Left (case 1)
Fig. 7 shows the comparison between active and passive
suspension for body displacement. Body displacement is used
to measure passenger ride comfort. Even it shows that the
amplitude of active suspension using LQR technique is higher
than passive suspension but it improved displacement settling
time.
Fig. 8 and Fig. 9 show the performance of wheel deflection
for each wheel. It clearly shows that the amplitude for active
suspension lower than passive suspension.
In the simulation results also shows that the output
performances for rear wheels give slightly higher output
compare to front wheels it may cause by output from rear
16
actuators force which is slightly higher compare with the front
actuators.
Fig. 12. Force Generate for Front Right and Left Actuator by Using LQR
controller (case 2)
Fig. 13. Force Generate for Rear Right and Left Actuator by Using LQR
controller (case 2)
By using same parameter setting for Q and R output
performance is captured to investigate either LQR can adapt in
changes of road disturbance.
Fig. 14. Body Displacement (case 2)
Fig. 15. Wheel Displacement for Front Right and Left (case 2)
Fig. 16. Wheel Displacement for Rear Right and Left (case 2)
Fig. 17. Suspension Travel Front Right and Left (case 2)
Fig. 18. Suspension Travel Rear Right and Left (case 2)
17
By comparing the performance of the passive and active
suspension system using LQR control technique for full car
model, it is clearly shows that there is a problem with
robustness. The output performances for case 2 happen to
have slightly higher amplitude compare with case 1 active
suspensions performances for certain parameters such as Body
Acceleration and Wheel Deflection for each wheel. Body
Displacement improved even the amplitude is slightly higher
compare with passive suspension system but the settling time
is very fast. Body Displacement is used to represent ride
quality. It is shows that LQR does not have capability to adapt
variations in road profiles.
VI. CONCLUSION AND FUTURE WORK
LQR control technique has successfully implemented to
the linear model active suspension system for full car model.
Active suspension system that has been proposed in [1] is for a
quarter car model. Input disturbance using in [1] is step input
and random input. In this study by using dynamic model in [1]
simulation has been done by using same road profile with full
car model to compare the performance of LQR between this
two model. Results show that there is a limitation using LQR
control technique. It cannot perform in rough road
disturbances especially for full car model. Performance shows
that LQR control technique for a quarter car model can
provide better ride comfort compare LQR control technique
that implement in a full car model active suspension. However
the LQR control technique still can improve ride comfort and
car handling compare to the passive suspension.
In this study also does not include dynamic models of the
actuators. Force actuator generates needed force to achieve
desired objectives. Thus the study of actuator must be
including in future work so it can give real time performance.
Different types of controller will be implementing to observe
the output performance of the suspension.
REFERENCES
[1] Y.M. Sam, M.R.A. Ghani and N. Ahmad, LQR controller for active car
suspension, IEEE Proceedings of TENCON 2000, 2000. I441-I444.
[2] Kim C., Ro P.I, An Accurate Full Car Ride Model Using Model
Reducing Techniques, Journal of Mechanical Design, vol. 124, pp.
697-705, December 2002.
[3] S. Ikenaga, F. L. Lewis, J. Campos and L. Davis, Active Suspension
Control of Ground Vehicle based on a Full-Vehicle Model,
Proceedings of the American Control Conferenc, Chicago, Illinois June
2000.
[4] Sam Y.M., Osman H.S.O., Ghani M.R.A., A Class of Proportional-
Integral sliding Mode Control with Application to Active Suspension
System System and Control Letters. 2004. 51:217-223
[5] Sam, Y.M. Proportional Integral Sliding Mode Control of an Active
Suspension System. Malaysia University of Technology. PHD
Dissertation. Malaysia University of Technology; 2004
18

Das könnte Ihnen auch gefallen