416 IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 19, NO. 2, MARCH 2011
Finite Frequency Control for Vehicle Active Suspension Systems
Weichao Sun, Huijun Gao, Senior Member, IEEE, and Okyay Kaynak, Fellow, IEEE
Abstract—This brief addresses the problem of control for
active vehicle suspension systems in finite frequency domain. The
performance is used to measure ride comfort so that more
general road disturbances can be considered. By using the gener-
alized Kalman–Yakubovich–Popov (KYP) lemma, the norm
from the disturbance to the controlled output is decreased in spe-
cific frequency band to improve the ride comfort. Compared with
the entire frequency approach, the finite frequency approach sup-
presses the vibration more effectively for the concerned frequency
range. In addition, the time-domain constraints, which represent
performance requirements for vehicle suspensions, are guaranteed
in the controller design. A state feedback controller is designed in
the framework of linear matrix inequality (LMI) optimization. A
quarter-car model with active suspension system is considered in
this brief and a numerical example is employed to illustrate the ef-
fectiveness of the proposed approach.
Index Terms—Active suspension systems, constraints, finite fre-
quency, generalized KYP lemma, control.
I. INTRODUCTION
A
VEHICLE suspension system basically consists of wish-
bone, spring, and shock absorber to transmit and filter
all forces between body and road. The spring is to carry the
body-mass and to isolate the body from road disturbances and
thus contributes to ride comfort. The task of the damper is the
damping of body and wheel oscillations, where the avoidance of
wheel oscillations directly refers to ride safety. Since the vehicle
suspension system is responsible for ride comfort and safety, it
plays an important role in modern vehicles.
In recent years, a lot of efforts have been made to develop
models for suspension systems and to define design specifica-
tions that reflect the main objectives to be taken into account.
In this sense, ride comfort, road-holding ability, suspension de-
flection, and actuator saturation are important factors to be ad-
dressed by any control scheme. However, these requirements
are conflicting. For example, increasing ride comfort results in
larger suspension stroke and smaller damping in the wheel-hop
mode. Therefore, vehicle suspension design requires a com-
promise between ride comfort and vehicle control. To achieve
Manuscript received July 24, 2009; revised January 09, 2010. Manuscript
received in final form January 28, 2010. First published March 11, 2010; cur-
rent version published February 23, 2011. Recommended by Associate Editor
C.-Y. Su. This work was supported in part by National Outstanding Youth Sci-
ence Fund (60825303), by 973 Project (2009CB320600), by the Heilongjiang
Outstanding Youth Science Fund (JC200809), and by the Foundation for the
Author of National Excellent Doctoral Dissertation of China (2007B4).
W. Sun and H. Gao are with the Space Control and Inertial Technology Re-
search Center, Harbin Institute of Technology, Harbin 150001, China (e-mail:
1984sunweichao@gmail.com; hjgao@hit.edu.cn).
O. Kaynak is with the Electrical and Electronic Engineering De-
partment, Bogazici University, Bebek 80815, Istanbul, Turkey (e-mail:
okyay.kaynak@boun.edu.tr).
Color versions of one or more of the figures in this brief are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCST.2010.2042296
a compromise between the performance requirements, a con-
siderable amount of research has been carried out for the last
few decades [3], [17], [21]. Among the proposed solutions, ac-
tive suspension is a possible way to improve suspension per-
formance and has attracted much attention [10], [19], [24], and
many active suspension control approaches are proposed, based
on various control techniques such as linear quadratic Gaussian
(LQG) control [4], adaptive control and nonlinear control [11],
fuzzy logic and neural network control [15], and control
[14]. In particular, active suspensions have been intensively
discussed in the context of robustness and disturbance attenua-
tion [6], [7]. Therefore, in recent years, more and more attention
has been devoted to the control of active suspensions, and
a number of important results have been reported, see for ex-
ample, [5], [13] and the references therein.
The most important objective for vehicle suspension systems
is the improvement of ride comfort. In other words, the main
task is to design the controller which can succeed in stabilizing
the vertical motion of the car body and isolating the force trans-
mitted to the passengers as well. In the literature it is possible
to find many results which aim at improving ride comfort [8],
[20], [22]. These results can effectively achieve desired vehicle
suspension performance, especially the ride comfort. It is worth
mentioning that most of the reported approaches are considered
in the entire frequency domain. However, active suspension sys-
tems may just belong to certain frequency band, and ride com-
fort is known to be frequency sensitive. From the ISO2361, the
human body is much sensitive to vibrations of 4–8 Hz in the ver-
tical direction. Hence, the development of control in finite
frequency domain is significative for active suspension systems.
The current approach for finite frequency domain is to
introduce the weighting functions. The weighting method is
useful in practice, however, the additional weights increase the
system complexity. Besides, the process of selecting appro-
priate weights is time-consuming, especially when the designer
has to shoot for a good tradeoff between the complexity of the
weights and the accuracy in capturing desired specifications.
An alternative approach is to grid the frequency axis. This ap-
proach has a practical significance especially when the system
is well damped and the frequency response is expected to be
smooth. But it lacks a rigorous performance guarantee in the
design process.
Another approach that avoids both weighting functions
and frequency gridding is to generalize the fundamental ma-
chinery, the Kalman–Yakuboviˇ c–Popov (KYP) lemma. The
KYP lemma establishes the equivalence between a frequency
domain inequality for a transfer function and a linear matrix
inequality (LMI) associated with its state-space realization
[1], [9], [12]. It allows us to characterize various properties of
dynamic systems in the frequency domain in terms of LMIs.
However, the standard KYP lemma is only applicable for the
infinite frequency range. Recently, a very significant devel-
1063-6536/$26.00 © 2010 IEEE