J Control Theory Appl 2013 11 (1) 80–85
DOI 10.1007/s11768-013-1010-9
Reliable H-infinity filtering for linear systems with
sensor saturation
Ting SHI
1†
, Hongye SU
2
, Jian CHU
2
1.Institute of Automation, Hangzhou Dianzi University, Hangzhou Zhejiang 310018, China;
2.State Key Laboratory of Industrial Control Technology, Institute of Cyber-Systems and Control,
Zhejiang University (Yuquan Campus), Hangzhou Zhejiang 310027, China
Abstract: In this paper, a new reliable H-infinity filter design problem is proposed for a class of continuous-time sys-
tems with sensor saturation and failures. Attention is focused on the analysis and synthesis problems of a full order reliable
H-infinity filter such that the filtering error dynamics is asymptotically stable with a guaranteed disturbance rejection atten-
uation level γ. It is shown that the filtering error dynamics obtained from the original system plus the filter can be modeled
by a linear system with sector bounded nonlinearity. The design conditions are given in terms of solutions to a set of linear
matrix inequalities (LMIs). These conditions are then considered in a convex optimization problem with LMIs constraints
in order to design an optimal reliable H-infinity filter. A numerical example is given to illustrate the effectiveness of the
proposed results.
Keywords: H-infinity filtering; Reliable filtering; Sensor saturation; Sensor failures; Linear matrix inequality
1 Introduction
It is well known that filtering problems play an impor-
tant role in signal processing. Various filtering schemes
have been presented recently, such as robust filtering [1],
Kalman filtering [2], H
∞
filtering [3–4], set-membership
filtering [5] and FIR filter [6]. In most existing literature,
it is commonly assumed that the sensors can provide un-
interrupted signal measurements. In the practical filtering
problem, the measurement constraints should be taken into
account. On the one hand, contingent failures are possible
for all sensors in real-world systems. A large degree of fil-
ter performance may degrade. In some safety critical sys-
tems, such as flight control and navigation systems, filter
schemes should ensure performances in spite of the pres-
ence sensor failures [7]. Following the general problem of
‘reliable’ control [8–12], a filter designed to tolerate sen-
sor failures while retaining desired performance is named
as ‘reliable’ filter in this paper. Relatively few attention
have been focused on the design of reliable filter systems in
the literature. In [13], an observer that can tolerate failures
of outputs is considered. However, only asymptotic track-
ing of states is achieved without any performance analysis.
In [14], an LMI-based robust filter design method is pre-
sented to solve reliable guaranteed error variance filtering
problem. By combining the LMI approach for H
∞
filter de-
sign and the adaptive method, a method of adaptive H
∞
fil-
tering problem is proposed in [7], where the filter parameter
matrices are updated automatically to compensate the sen-
sor failures effects on systems. However, as it is also pointed
out in [7], this method makes the computation very costly.
On the other hand, in practical control systems, sensors
and actuators cannot provide unlimited amplitude signal
due primarily to the physical, safety or technological con-
straints [12]. Because of their theoretical significance and
practical importance, the problem of controller design for
systems with actuator saturation have been extensively in-
vestigated, see, e.g., [15–19]. Comparing with the vast lit-
erature with respect to control problem for systems with ac-
tuator saturation, the filtering results for sensor saturation
have been relatively few probably because of the technical
difficulty, see [5, 20].
In practical systems, sensor failures and saturations are
both frequently encountered. Unfortunately, to the best of
the author’s knowledge, the filtering problem for systems
subject to sensor failures and saturation simultaneously is
still an open problem that deserves further investigation. It
is therefor, the objective of this paper to fill the gap. In this
paper, we aim to investigate the H
∞
filtering problem for
systems with both the sensor failures and saturation. Note
that the systems model addressed is quite comprehensive to
cover sensor failures and saturations, hence reflecting the re-
ality closely. The main contribution in this paper is that our
method can deal with the sensor subject to both the failures
and saturations while the existing results cannot be imple-
mented directly.
Notation The notation is standard. For real symmet-
ric matrices X and Y , the notation X>Y(respectively
X Y ) means that the matrix X − Y is positive-definite
(respectively, positive semidefinite). I is the identity ma-
trix with appropriate dimensions. ·refers to either the
Euclidean vector norm or the induced matrix 2-norm. The
notation L
2
[0, ∞) represents the space of square-integrable
Received 10 January 2011; revised 1 August 2011.
†
Corresponding author. E-mail: tingshi@hdu.edu.cn.
This work was supported by the National Creative Research Groups Science Foundation of China (No. 60721062), the Foundation of Zhejiang
Department of Education (No. Y201121463), and National Natural Science Foundation of China (Nos. 61203025, 60974138).
c
South China University of Technology and Academy of Mathematics and Systems Science, CAS and Springer-Verlag Berlin Heidelberg 2013