See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/4253937

Semi-active Suspension System Design for Quarter-car Model using Model

Reference Sliding Mode Control

Conference Paper · January 2007

DOI: 10.1109/ICVES.2006.371623 · Source: IEEE Xplore

CITATIONS READS

18 1,313

2 authors:

Jialing Yao Jia-qiang Zheng

Nanjing Forestry University Nanjing Forestry University
14 PUBLICATIONS 153 CITATIONS 4 PUBLICATIONS 83 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

suspension View project

All content following this page was uploaded by Jialing Yao on 01 August 2015.

The user has requested enhancement of the downloaded file.

 Semi-active Suspension System Design for
Quarter-car Model using Model Reference
Sliding Mode Control
Jia-ling Yao, Jia-qiang Zheng

One of the most popular and implemented controllers in

Abstract- This paper investigates the control of a semi-active commercial applications is the skyhook damping concept [3].
suspension system using a quarter-car model, and proposes a The practical implementation calls for the use of an actuator
model reference sliding mode controller which is easy to be carried between the sprung and the unsprung masses, because finding
out and eliminates the necessity of a road signal as well as an imaginary point in the sky for fixing the damper is not
measuring damper force. The proposed controller utilizes an
approximate ideal skyhook system as a reference model, and the
control law is determined so that an asymptotically stable sliding mass dynamic performances as the controller force input has to
mode will occur in the error dynamics between the plant and the be applied on both the sprung as well as the unsprung masses.
reference model states. In order to improve the dynamic quality of Thus the dynamic response of the practical skyhook damping
sliding mode motion, a uniform reaching law is used. Numerical system is considerably worse than that of the ideal
simulation results suggest that the proposed controller can achieve skyhook-based suspension system [4].
a high degree of robustness against model uncertainties and Since actual suspension system contains many uncertainties,
disturbances. It obtains much better ride quality than that of including the system non-linearities, the passivity constraint and
passive and practical skyhook case, and achieves better handling the modeling errors, a more robust control scheme is required. A
performance over other controllers. sliding mode controller (SMC), which is suitable for linear
systems as well as non-linear ones, is convenient for practical
application, and can achieve high robustness against model
I. INTRODUCTION uncertainties and disturbances [5]. An important issue for
V ehicle semi-active suspension systems are less complex, developing a sliding mode controller is to maintain the sliding
more reliable and safer, and don't require an external mode in the plant dynamics. An alternative way of using the
power source. They can lower the vibration transmission nearly sliding mode controller is that one can use it in a force feedback
as much as fully active suspension systems [1]. Thus loop, which is incorporated to induce the damper to generate the
semi-active suspension system becomes a hot spot for research desired force [6]. However, this way requires the measurement
and commercial development worldwide now. Many kinds of of the damper force, and it is expensive and impractical for
controllable dampers with viscous/viscoelastiplastic fluids such mass-production cars.
as magnetorheological (MR) fluids have been developed, and The objective of this study is to design a model reference
have already been commercialized. This makes semi-active sliding mode controller (MRSMC) which will easily be carried
suspension systems have more ideal controllable dampers. out. According to the theory of model reference control (MRC),
Another key problem in the implementation of the semi-active this controller will utilize a modified ideal skyhook damper
suspension systems is the design of an appropriate system as a reference model for eliminating the necessity of a
implementable algorithm that can satisfy the conflicting road signal as well as measuring damper force, which is trying
requirements of achieving more driver comfort as well as better to make the whole process simple and easy to implement, and
vehicle handling performance. At present, there are so many result in better ride quality and handling performance. The
theoretical study methods, including adopting the up-to-date effectiveness of the controller is investigated by numerical
research achievements of the modern control theory, however, simulation.
there are fewer methods which have been put into service. The
emphases of research focus on how to develop a practical, II. MODELING OF THE SUSPENSIN SYSTEM AND
reliable and lower cost control strategies [2]. PROBLEMS
A. Quarter-car vehicle model
Jia-ling Yao is with the College of Mechanical and Electronic Engineering,
Nanjing Forestry University, Jiangsu 210037, China (phone: A quarter-car model of the suspension is a two-DOF (degree
0086-25-85428637; e-mail: [email protected]). of freedom) model, which models the vertical or the heave
Jia-qiang Zheng iS now with the College of Mechanical and Electronic
Engineering, Nanjing Forestry University, Jiangsu 210037, China (e-mail motion of the vehicle alone. As the design goal of most of the
[email protected]). semi-active suspension is to reduce the vertical acceleration, the

1-4244-0759-1/06/$20.OO ©2006 IEEE. - 398 -

quarter-car model is sufficient from the controller design point approximation, the measured states of the unsprung mass can be
of view [7]. Hence the quarter-car system is chosen as the used directly as input to the reference model. However, the
vehicle model in this paper for the MRSMC design, as shown in reference model is not an ideal skyhook system but an
the left part in Fig. 1. In general, the damper force of the tyre is approximate one with an ideal controllable damper. The
neglected because it is very tiny. advantages of the reference model are: (1) it has an ideal
controllable damper, (2) the reference model is under the same
constraints as those of the plant, (3) it will not lead to
deterioration of the unsprung mass dynamic performance.
mJGmJ 4t The reference model dynamic equation can also be derived
from Fig. 1.
kCit k
cc.R''; m sr =
f ms -ks(Xsr - x) -c(Xs - i) - cxsr
A
= xa AThe state variable forms are as follows:
xrl = sr ' xr2 Xsr
kRt d efrrence The state equation is
Xr = ArXr+ Brux
Planit 3
Fig. I Semi-active suspension system using reference model r
X=[i]
[
r

'
u=[]
x

The plant dynamic equations can be derived from Fig. 1. where

{Cs fo r x2(X2-X4) >O
co =
m xs =-ks (Xs-x )-Cc (s x-
)-ufd {
for
x2(X2 -
X4) < 0

muXu =ks (xs -x)u + cc (.5-iu)-kt (xu -d) + fd Reference

The state variable forms of the equations can be obtained as Road input modef
follows:
xI =X ' x2=x s x3 =x , x4 =xu I x5 =x -d Fig.2 Model reference sliding mode control system
The state equation is
X = AX +Bfd +Fd C. Sliding mode controller
fd =F(X2-X4,i) The proposed controller makes the plant sprung mass track
where fd is the damper force which is a nonlinear function the motion of the reference model sprung mass, and forces the
tracking error dynamics between the plant and the reference
of the control input current i and the piston speed of the model states in an asymptotically stable sliding mode.
damper x2 - According to the equations of motion of the plant and the
reference model, define a tracking error vector
B. Rreference model e = Tl-X1 X2-X2]
Model reference control (MRC) operates on the basic Fig.2 shows the block diagram of the model reference sliding
principle of making a given system behave as a desired system mode control system. Then the tracking error dynamics is
by the application of a suitable control force. The desired follows
system is referred to as the model reference system. The A ee+H X + B f
versatility of the MRC method lies in the fact that the model e eXr efd
reference system does not need to be a practical system, but can O 1 O
be any ideal mathematical model, and does not need to be here A k c B
practically feasible [8], however, it should be similar to the e s c e
practical system as far as possible. In a typical implementation Ms Ms Ms
of the model reference control, system input is provided to the O O
reference model, and the difference of responses between the
system and the reference model is compensated by a controller. He = Cc
However, in the case of a suspension system, it is very difficult
to measure the road height signal for the reference model. In this
study, a simplified reference model is used to solved this Then, a surface of the error vector is defined,
problem, as shown in the right part of Fig.l1. Since the tyre is s= ce
almost ten times stiffer than the suspension spring, in general, c = [cl 1]
the motion of the unsprung mass can serve as a good If the error dynamics is in sliding mode, that is s = s= 0.
approximation to the road input within the normal operating Using this expression, the equivalent force can be found
frequency range of a suspension system. Using this

- 399 -
 fde -scmdeq = Il-k5c
c~~~~~~1m5 -ccl + CoX2 ]e
coefficient of the passive system. The parameter values used in
the simulation are as follows:
In order to improve the dynamic quality of sliding mode =16k, =20g k=I00Nm
motion, an uniform reaching law is used, = 16k,m2Og 100Nm
ds kt =100000N/m, c = 2000 Ns/m, cc=126Ns/m,
dt -
The desired force which is constrained by passivity is as
)cp =lIlOONs/m , r =80, csh = 2000Ns/m
The simulation is carried out with B grade road model as the
follows:I random input. The road roughness coefficient is
ffd fdeq + ysgn(s), for (fdeq + ysgn(s) )(2 -XI) > 0
0 for (fdeq +ysgn(s))(x2 X4) < 0Gqn,=6xI0-m,tevhceeliyisv6k/hte
Gqn)6

frequency no =0. 1m-1,

0

sampling
of reference space is
n,tevhcl

the
eoiyi Omh h

III. SIMULATION RESULTS period for control is t =0. 0 1 s, and the power spectral density
In order to investigate the performance of the sliding mode (p.s.d.) of a band-limited white noise input is
controller, a numerical simulation is performed using Gq(f ) = 4Zt2Gq(nfln )02 V.-
MATLAB7.0/Simulink6.0, and the time domain and frequency Comparing the practical skyhook system with the reference
domain numerical simulations of the passive suspension, the model through adopting the input way shown in Fig. 1, this
practical skyhook, the ideal skyhook, and the MRSMC-based controller can not achieve the performance of the ideal
controller are carried out. The control scheme proposed bY approximate skyhook system at all, as there is not a feedback to
Karnopp [9] for the model of the practical skyhook system (left compensate the damper force error. The simulation is shown in
part in Fig. 1) is a kind of clipped-optimal control. The damper Fig.3. Meanwhile, under the same condition, the model
force is decided by reference sliding mode controller yields almost perfect tracking
{ck~ for kjk5'~s- ~)> 0 to the reference model, as shown in Fig.4. This means the
fd = f ) 0control scheme is highly robust against the damper force error.

00~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~00

E 0

0.06 Z
-- --
6 7 8 9
T- -m- --[-
1

2 3---------------------
4--5 ---------------- Sprung---I--mass-- d-----Isplacement----------I-----
10(a)-----
Time [sec] ~~~-.2-----------------
(a)i----- ---- Sprung ----mass displacem en ------------I-----

--- -- ime--- [- -- --ec]-

0.02 -------- I-------- ------------------- r---b---Sprung--m ass-velocity-
Fig 5(a), obtained from---th Imedoai-smuatonan
are~~~~~~~~~~~~~~~~~~-.0
(b) Time [sec]
(b)----- SpugmasvloiyFi6a-i-baie-rm-h-reuny-oan-iuato.I
0.03 canIe-fundtha-th-spungmassresons-isconideabl
---- -----------

Fig.3 Practical skyhook controller the ----MRSM case- as-compared-to-the-passiv

better~~~~~~~~~~~~~~~~~~~0.0for-----------------
Themoelofth iealskhok ysemis hon n heriht susenio rspns fo te ccleaton-as,-ndth-MSM
part4 inFi.1,an-hesetonude-teunprnIms-i-te repos-i-as-mc-bttrthn ht-fth-patialsyho
same as----th-oe-nerth -ef-pr in-ig 1.- CIhisthedapin-rspose
co -0~ ~ ~ ~ ~ ~ ~ae as expected,--and-tends-to-that-of the----I--ideal----skyhook------------
coefficient of-----the----
ida- khoksse.--i-h-apig Fo-is-() d f hiedoansmltinadFg

40.07
 6(b), (c) of the frequency domain simulation, it can be found
200 .------ that in the suspension deflection and the tyre deflection cases
the MRSMC response is better than the practical skyhook case,
100
and is nearly the same as the ideal skyhook case, but is worse
Tethan the passive case.

20 M0SMC
-20 ------
0.80 P
PracticalSkyhook
300 ) 0.16 Ide Skhook
0 5 1 2 53 4 5 6 7 8 4 10
Time [sec] 0.14

(c) Damper force 0.12-

Fig(4 Model reference sliding mode controller 1.
0.08

_____ MRSMC 0.06 -

-1-3 5 L r | n | r 1 yE P0ssive 0.0

----------- .0 |-

(- 1- 15 5- 10 1 10

Frequency [Hz]
(a) Sprung mass acceleration p.s.d.
-1 --- ~~~~~~~~~~~xiS-[rr/H -z]

125------------ ----'--- ---------- ---',------ --- ------ -------

----------
a -
c oS k
t-- ----------- 0.04 .\ |
--r -----
------------ --------- Passive
\g|
Practica2.5
lS-kyhook
~~~~~~~~~~~~~~~~MPSMC
Fig.6 Frequency domain simulation
00-b - l, - bIdelSkyhook

0 0.5 1 0 15 2 25 3 35 4 45 5 2
Time [sec]
(a) Sprung mass acceleration 1.5
MRSMC
01 - ---------------------------------o--------hPractical Skyhook 1-

0.5~~~~~~~~~P0
0~~~~~~~~~~~J]1 Frequency[Hz]1

oo k -7 s(b) Suspension deflection p.s.d.

oyA tt 1 \ I i ! % It A\21 t t changed, thatis, decreasing the sprung mass[r -i] adnei
PSD
__1_
O 05 1 1.5 2 1 5 3
P
2 1- vtical Skyhook
IdealSkyhook
3
Time[sec]
(b) Sprung mass acceleration 2.5 pI

0 1.5-

0.005
0 ~~~~~~~~~~~~0.5
-~-

10o 1011
-0.01 Fre~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~[Hz]~Pquecy
-0015 ___MSC(c) Tyre deflection p.s.d.
-0.02 --Passive Fig.6 Frequency domain simulation
Practical Skyhook
010.2 0 0.3 0.4 Time0-5[sec] 0.6 0.7 0.8 0.9 odr ivsiae
In odrto ivsgaethe routesof temdlreference
rbsns h oe

(c) Suspension deflection sliding mode controller for the plant parameter perturbations
0.01 5
~~~~~~~~~~~~and
disturbances, comparing with the passive suspension, the
0.01 ~~~~~~~~~~~~practical skyhook, and the ideal skyhook systems in the power
0.005 ~~~~~~~~~~~~spectral density (p.s.d.), some parameters of the plant are
changed, that is, decreasing the sprung mass 10% and increasing
ID d
~ ~ ~~K ithe springy stiffness 20%, then m< =144kg
which the practical skyhook controller yields is much greater
than the previous case. From this result it can be concluded that
the model reference sliding mode controller is highly robust
against model uncertainties and disturbances.

[(m/sec') / H1Z] PSD

_ MRSMC
0.16 -------- Passive
Practical Skyhook
0.14 -- Ideal Skyhook

0.12

0.1 / >

0.08/
00.060 _-g- 7 / \ , /_

0.04

002_

0
10 10 10
Frequency [Hz]

Fig.7 Sprung mass acceleration p.s.d. for

plant parameter perturbations

IV. CONCLUSIONS
This paper proposes a model reference sliding mode
controller which is easy to be implemented and eliminates the
necessity of a road signal as well as measuring damper force.
Comparison of the performance of the proposed controller with
the passive suspension, the practical skyhook and the ideal
skyhook systems is carried out. Numerical simulation results
show that both ride quality and handling performance are
improved using the MRSMC, and the ride quality using the
MRSMC tends to that using the ideal skyhook system, however,
the handling performance is worse than the passive case. This
approach towards the design of semi-active suspensions can be
further pursued for a practical set-up that could validate the
theoretical predictions and the simulation results.

REFERENCES
[1] Song Xubin, Ahmadian Mehdi, Southward Steve, "An Adaptive Semiactive
Control Algorithm for Vehicle Suspension Systems". Proceedings of ASME
IMECE 2003: 2003 International Mechanical Engineers Conference: 1 - 9.
[2] Yao Jialing, Cai Weiyi and Chen Ning, "A Review on the Development
Status of Automotive Semi-active Suspension Systems", Automotive
Engineering, 28(3), 276-280, 2006
[3] Nagai, M. "Recent researches on active suspension for ground vehicles".
Jap. Soc. Mech. Engrs Int. J. Ser. C, 1993, 36(2), 161 - 170.
[4] Kamopp, D. "Active damping in road vehicle suspension systems". Veh.
System Dynamics, 1983, 12, 291 - 316.
[5] ZHANG Xiao-yu, SU Hong-ye. "Survey on the Developments of Sliding
Mode Variable Structure Control Theory". CONTROL AND
INSTRUMENTS IN CHEMICAL INDUSTRY, 2006, Vol.33, No.2 p1 -8.
[6] S.J. Dyke, B.F. Spencer Jr., M.K. Sain and J.D. Carlson, "Seismic
Response Reduction Using Magneto-rheological Dampers", Proc. of the
IFAC World Congress, L, 145-150, 1996.
[7] Hrovat, D. "Survey of advanced suspension developments and related
optimal control applications". Automatica, 1997, 33(10), 1781 - 1817.
[8] Sunwoo, M. and Cheok, K.C. "Model reference adaptive control for
vehicle active suspension systems". IEEE Trans. Ind. Electronics, 1991,
38(3), 217 - 222.
[9] D. Karnopp, M. J. Crosby and R.A. Harwood, "Vibration Control Using
Semi-Active Force Generators", J. of Engineering for Industry, 619-626,
1974.

- 402 -
View publication stats