Actuator Power Consumption of Active Suspension System with

Override Control Strategy

Unggul Wasiwitono
Mechanical Engineering Department, Faculty of Industrial Technology, Institut Teknologi Sepuluh
Nopember (ITS), Kampus ITS, Surabaya, 60111
unggul@me.its.ac.id

Keywords: Active suspension, Actuator saturation, Suspension deflection, Override control, Power
consumption.

Abstract. The main function of the vehicle suspension is to improve ride and handling performance.
In vehicle active suspension, better ride comfort is usually required larger control input and larger
suspension deflection. However, the actuator that deliver the control signal have a limitation which
is commonly known as actuator saturation. There is also a structural constraint that limits the
suspension deflection. In this study, an alternative approach to the vehicle active suspension system
is proposed. In this approach, some separation in the controller such that one part is devoted to achieve
nominal performance and the other part is devoted to constraint handling is performed. In addition,
the actuator power consumption of the proposed control strategy is further investigated numerically.
The simulation results show that the proposed control strategy can manage the trade-off between
performance and the actuator power consumption.

Introduction
More and more systems are involved in today cars to guarantee both safety (e.g. ABS, ESP, etc.)
and comfort (e.g. suspension control, cabin noise control, etc.). Concerning the vehicle comfort and
road holding ability, the suspension system has the major roles to provide 1) isolating passengers
from vibration and shock arising from road roughness (ride comfort); 2) suppressing the hop of the
wheels so as to maintain firm and uninterrupted contact of wheels to road (good handling or good
road holding); and 3) keeping suspension deflection within an allowable maximum [1]. The most
important objective for the vehicle suspension system is the improvement of the ride comfort [2] and
a significant control input is often necessary to obtain better performance. However, in practice, the
actuators which deliver the control signal are always subject to limits in their magnitude which is
commonly known as actuator saturation. In addition, there is also a structural constraint that limit the
suspension deflection.
There are several ways of dealing with the control and output constraints. One can use the control
and output constraints as optimization constraints such as used in [1, 2]. This leads to a quite
significant linear programming problem and a controller can be designed such that it ensures that the
state always belongs to the maximum output admissible set. The finite frequency control has
been proposed in [2] which shows that an improvement in ride comfort is obtained compared with
the entire frequency control. Although the aforementioned controllers have been designed so as
to maintain the suspension deflection in a certain range based on a regulated road disturbance, the
control input and the suspension deflection might reach the limitation due to various road shapes. One
way of incorporating output constraints into controller design is using Model Predictive Control
(MPC) strategies. However, the MPC approach is generally expensive in terms of computation [3].
The other way of dealing with the output constraints is override control [3, 4, 5], and in this study,
we consider the override control to tackle the problem of the output constraints.
In [6] an alternative approach for the vehicle active suspension system is proposed. A separation
in the controller such that one part is devoted to achieve nominal performance and the other part is
devoted to constraint handling is performed. This is the approach taken in anti-windup compensation
[7]. It is shown that the proposed control approach has the potential benefit in achieving the best
possible ride comfort. However, the output constraint due to suspension deflection limitation is not
considered. In this study, based on the two-step design procedure, the combination of override control
and anti-windup control strategy are considered to deal with output constraint and control input
constraint. Local control design technique based on the circle criterion and 2 gain performance is
used for the override and anti-windup compensator synthesis. In addition, the actuator power
consumption of the proposed control strategy is analyzed.

Problem Formulation
The quarter-car model is very often used for the vehicle suspension analysis and design, because
of its simplicity yet capture many important characteristics of the full model. Define, is the sprung
mass, which represents the car chassis; is the unsprung mass, which represents mass of wheel
assembly; and are damping and stiffness of the suspension system, respectively; and stand
for compressibility and damping of the pneumatic tire, respectively; and are the displacements
of the sprung and unsprung masses, respectively; is the road displacement input; and u is the active
input of the suspension system. Then, define the following state variables:

1 = , 2 = , 3 = , 4 = (1)

where 1 denotes the suspension deflection, 2 denotes the tire deflection, 3 denotes the sprung
mass velocity, and 4 denotes the unsprung mass velocity. Further, define the disturbance input as
= , and = [1 2 3 4 ] , then, by applying Newtons second law of motion and using the
static equilibrium position as the origin, the state-space form of the vehicle suspension system can be
written as

= + + (2)

where is the suspension system state, is the disturbance input, is the

control input, and matrices , and are appropriately defined.

(a) (b)
Figure 1. Override control strategy

In this study the application of the override and anti-windup control strategy for the vehicle active
suspension as shown in Fig. 1a is studied. In this Figure is the constrained output, and is
deadzone function to control the activation of the compensator . Further, by defining

0
= [ ], = [ ] (3)
0

The closed-loop system shown in Fig. 1a can be represented as that shown in Fig. 1b, with the plant
P is described as
 x p Ap x p Bu u Bd d
z1 Cz1 x p Dzu1u Dzd1d (4)
yl C yl x p Dyu u

Assume that the linear control K has been designed, let now focused on the controlled output for
the design of the override and anti-windup compensator . The most important objective of the active
vehicle suspension is the improvement of the ride comfort, hence, the sprung mass acceleration is
chosen as performance output and we have

1
1 = [

0
], 1 = [ ], 1 = [0] (5)

Furthermore, because the control objective of the override and anti-windup control is to avoid the
suspension reach its stroke limitation and preventing the actuator from saturation, beside the
aforementioned performance objective, we want to minimize 2 = therefore, we will consider the
next performance output

1 1
=[ ] (6)
2 2

The constant weights 1 and 2 are used for the tuning of the balance between the ride comfort
requirements and the minimization of 2 . Thus the problem is to minimize subject to

2 < 2 (7)

where is defined by Eq. (6) with internal stability.

Override and Anti-Windup Compensator Synthesis

A state feedback control system is considered in this study and therefore, the control signal can
be described as.

= (8)

Define

1 1 0
= [ 1 1 ] , = [ ], = [ 1 1 ] , = [ ] (9)
0 0 0 2

then, the system in Fig. 1b can be represented as that of Fig. 2, with the system G is described by

x p Ap Bu K x p Bu v Bd d
z Cz Dzu K x p Dzu Dzv v Dzd d (10)
yl C yl Dyu K x p Dyu v

Further, by considering that () satisfies the sector condition [8] in the finite interval [ , ] with
1
= (1) , then the following inequality condition holds

( ) 0 (11)
 Figure 2. Nonlinear feedback system with sector condition

and the next theorem guarantees the 2 gain condition Eq. (7) for the nonlinearity that satisfies Eq.
(11).

Theorem 1. For a given , if there exist a positive-definite symmetric matrix , a diagonal

matrix = [1 , 2 , , + ] > 0, and a scalar > 0, that satisfies the next matrix inequality

+ + + + +

+

+ + 2

0

<0
0

[ + ]
(12)

Then, the feedback system shown in Fig. 2 with compensator = 1 is asymptotically stable and
1
the 2 gain from to is less then when the condition | | (1 ) () (), = 1, 2, ,

holds.
Proof. Omitted due to page restriction.

This design problem is an LMI problem with respect to the variables , , , , and therefore it can
be solved easily by a numerical optimization method.

Numerical Example
We assume that the linear controller K has been designed, and for the current study we consider
the state feedback controller K reported in [2]. Fig. 3 shows the frequency response from the
disturbance to the body acceleration and suspension deflection for the case of passive suspension and
active suspension with state feedback control. It can be clearly seen that this state feedback controller
yields the lower value of norm over the frequency range 1 8 Hz compared with the passive
system. However, in term of suspension deflection this state feedback controller yields larger value
especially at low frequencies region.
Let us now consider the case of an isolated bump in a road surface. The corresponding disturbance
input is given by [2]

1
(1 cos(2)) 0
2
={ 1 (13)
0 <

where represents the amplitude of the bump and is disturbance frequency. Fig. 4 shows the
plots of body acceleration, suspension deflection and control input for the case of disturbance
frequency 1 Hz and bump amplitude 0.16 m. It can be seen in this Figure that the ride comfort by
state feedback control [2] is deteriorated caused by the suspension hits it limits, at around 0.5 s.
 Figure 3. Frequency response from disturbance to body acceleration and suspension deflection
(dotted line: passive suspension; solid line: state feedback control [2])

Figure 4. Bump response for the case of disturbance frequency 1 Hz and amplitude 0.16 m (dotted
line: passive suspension; solid line: state feedback control [2]); dashed line: proposed control
strategy)

To overcome this problem, we add override and anti-windup compensator and based on Theorem
1, we synthesize the compensator by setting = 1, 1 = 250 and 2 = 50, and obtain the following
result

= [2.55657 18.72049] (14)

Since the control input saturation is 1.5 kN and suspension deflection limit is 0.1 m, we choose
the deadzone limit = 1 kN and = 0.05 m, respectively. It can be seen from Figure 4 that
the improvement in ride comfort is obtained. The suspension deflection and control input is reduced
to be below the allowable bound.
Beside the aforementioned active suspension performance, the actuator power consumption is also
an important issue in active suspension control. In this study, the average power consumption is
calculated using the following formulation [2].

1
= 0 () () (15)

where () respects the displacement of actuator, and according to the installed location, the
displacement of actuator is equivalent to that of the suspension deflection and is the integral time.
Figure 5 shows the actuator power consumption of the proposed control strategy over the whole
frequency range of interest. It can be clearly seen from Fig. 5 that the proposed override control
strategy gives the lower actuator power consumption compared with the state-feedback control. This
shows that the proposed control strategy has the potential benefit in managing the trade-off between
performance and actuator power consumption.

Figure 5. Actuator power consumption (solid line: state feedback control [2]); dashed line: proposed
control strategy)

Summary
The application of the override control for active vehicle suspension system is studied. It is shown
that the occurrence of the suspension hits it deflection limits and control input reaches the saturation
bound can be minimized by the proposed control strategy. Furthermore, Simulation result shows that
the vehicle active suspension with the proposed control strategy yields the low value of the actuator
power consumption. This results show that the proposed control strategy has the potential benefit in
managing the trade-off between performance and actuator power consumption.

References
[1] H. Chen and K.-H. Guo, Constrained control of active suspensions: An LMI approach,
IEEE Transactions on Control Systems Technology, vol. 13, no. 3, pp. 412421, 2005.
[2] W. Sun, H. Gao, and O. Kaynak, Finite frequency control for vehicle active suspension
systems, IEEE Transactions on Control Systems Technology, vol. 19, no. 2, pp. 416422, 2011.
[3] M. C. Turner and I. Postlethwaite, Output violation compensation for systems with output
constraints, IEEE Transactions on Automatic Control, vol. 47, no. 9, pp. 15401546, 2002.
[4] M. C. Turner and I. Postlethwaite, Multivariable override control for system with output and
state constraints, International Journal of Robust and Nonlinear Control, vol. 14, pp. 11051131,
2004.
[5] Y. K. Park and H. Y. Young, Dynamic anti-windup based control method for state constrained
systems, Automatica, vol. 39, pp. 19511922, 2003.
[6] U. Wasiwitono and M. Saeki, Fixed-order output feedback control with anti-windup
compensation or active suspension systems, Journal of System Design and Dynamics, vol. 2, no. 2,
pp. 264278, 2011.
[7] S. Tarbouriech and M. Turner, Anti-windup design: an overview of some recent advances and
open problems, Control Theory & Applications, vol. 1, pp. 119, 2009.
[8] H. K. Khalil, Nonlinear Systems. Prentice-Hall, Inc, second ed., 1996.