Interval observer design for discrete-time
switched system
⋆
Shenghui Guo
1,2
Fanglai Zhu
1
1. College of Electronics and Information Engineering, Tongji
University, Shanghai 201804 China. (e-mail:
zhufanglai@tongji.edu.cn)
2. College of Electronics and Information Engineering, Suzhou
University of Science and Technology, Suzhou 215009 China. (e-mail:
12gsh@tongji.edu.cn).
Abstract: In this paper, the problems of interval observer design for discrete-time switched
system are investigated. In order to make the residual system stable and non-negative, a time-
varying linear transformation and a standard switched system stability analysis are used. Based
on these, an interval observer for discrete-time switched system is constructed. A simulation
example is given to illustrate the effectiveness and correctness of the proposed method.
Keywords: State estimation, switched system, discrete-time systems, interval observer,
stability analysis.
1. INTRODUCTION
For state estimation purpose, interval observers have b e-
come effective alternatives of traditional observers espe-
cially when systems suffer from uncertainties (Gouz´e et al.
(2000); Wang et al. (2015)). Different from traditional
observers such as asymptotical and H-infinity observers
which can produce state estimates in the sense of asymp-
totical or H-infinity convergence, interval observers pro-
vide the estimates of the upper and lower boundaries of
the system states. It can be used to resolve many prob-
lems that traditional observers can not. Up to now, many
effective design methods are proposed for both continuous-
time and discrete-time, standard and descriptor system-
s (Cacace et al. (2015); Efimov et al. (2013b,a, 2015);
Mazenc and Bernard (2011); Mazenc et al. (2014); Ra¨ıssi
et al. (2012); Thabet et al. (2014); Zheng et al. (2016b,a)).
For example, Efimov et al. developed interval observers
for both continuous and discrete time systems based on
time-invariant linear transformation and Sylvester equa-
tion (Ra¨ıssi et al. (2012); Efimov et al. (2013b)), while
Mazenc et al. construct interval observers by using time-
varying linear transformation and Jordan canonical form
(Mazenc and Bernard (2011); Mazenc et al. (2014)). Ca-
cace et al. proposed an interval observer design method via
the Internal Positive Representations of systems (Cacace
et al. (2015)). However, to the best of our knowledge, few
results of interval observer designs for switched systems
under average dwell time (ADT) have been reported in
literature (He and Xie. (2016)).
Actually, as a class of special hybrid systems, switched
systems can be used to describe a wide range of real
physical and practical systems including power electronics,
⋆
This work was supported by the National Natural Sci-
ence Foundation of China (61573256). Corresponding author: F.
Zhu(zhufanglai@tongji.edu.cn)
flight control systems, network control systems, and so
on, so they have been extensively studied in the past
decades(Daafouz et al. (2002); Zhang and Shi (2008);
Zhao et al. (2012)). In the past, much effort has been
made on the stability analysis under restricted switching
laws(Lin and Antsaklis (2009)), such as dwell time and
average dwell time (Zhang and Shi (2008); Zhao et al.
(2012)), and in the designs, usually multiple Lyapunov
functions (Branicky (1998); DeCarlo et al. (2000)), or
piecewise quadratic Lyapunov functions (DeCarlo et al.
(2000)) are considered. Little attention has b een put on
state estimation (Xiang et al. (2012); Zhao et al. (2015)),
especially the interval state estimation.
For all of the above reasons, in this paper, we will focus
on the interval observer designs for switched systems under
ADT. The paper is organized as follows. Section 2 gives the
problem statements and summarizes some preliminaries of
interval observers and switched systems. Section 3 presents
the main results of our work. A numerical example is given
in Section 4 to illustrate the effective of proposed method.
At last, some conclusions are summarized in Section 5.
Notation. Most of the notations used in the paper is
standard. R is the real set, R
n
is the n-dimensional
Euclidean space, and R
m×n
is the set of m × n real
matrices. N is the natural number set. Z
+
denotes the
set of non-negative integers, and R
n
+
is the set of n-
dimensional Euclidean space with non-negative elements.
For two vectors x, y ∈ R
n
and matrices A, B ∈ R
m×n
,
the relation x ≥ y (x > y) and A ≥ B(A > B)
are understood elementwise. For a matrix A ∈ R
m×n
,
define A
+
= max{0, A}, A
−
= max{0, −A}, then we
have A
+
≥ 0, A
−
≥ 0, and A = A
+
− A
−
. For a
square matrix A ∈ R
n×n
, A ≺ 0 means that A is a
symmetric negative definite matrix. C
1
denotes the space
of continuously differentiable functions, and a function
Preprints of the 20th World Congress
The International Federation of Automatic Control
Toulouse, France, July 9-14, 2017
Copyright by the
International Federation of Automatic Control (IFAC)
5229