Observer Design for Lipschitz Nonlinear Systems with
Nonuniformly Sampled Measurements
CHEN Wu-Hua
1,2
, WANG Zipeng, WEI Dan, LU Xiaomei
1. College of Mathematics and Information Science, Guangxi University, Nanning 530004, P. R. China
E-mail: wuhua
chen@163.com
2. School of mathematics and statistics, Hubei Normal University, Huangshi 435002, P. R. China
Abstract: This paper studies the design of full-order observer for Lipschitz nonlinear systems with nonuniformly sampled
measurements. By introducing a time-varying Lyapunov functional to capture the dynamic characteristic of the estimation error
dynamics, a less conservative condition has been found that guarantee the exponential stability of the estimation error dynamics.
The new condition is dependent on the upper bound of sampled intervals and is formulated in the form of linear matrix inequalities
(LMIs). The observer gain matrix can be achieved by solving a set of LMIs. Finally, two examples are given to illustrate the
feasibility and effectiveness of our results.
Key Words: Observer, Lipschitz Nonlinear Systems, Nonuniformly Sampled Measurements
1 Introduction
Observer design for nonlinear systems has received more
and more attention in the literature. This is because state
observation is one of the basic problems in control of non-
linear systems and has been found wide application in many
fields such as output feedback control, fault detection and
isolation, fault tolerant control, adaptive control, and secure
communication. For these reasons, various observer design
methods have been developed for different types of nonlinear
models, such as Luenberger observer [1], Lipschitz observer
[2], adaptive observer [3], partial-state observer [4], sliding-
mode observer [5], and so on. We noticed that all these ob-
server design techniques are based on the assumption that
the output can be measured continuously. But in the net-
work environment, the system output information is through
a shared digital communication network which is available
only at discrete time instants. Therefore, how to use the sam-
pling system information to reconstruct the system state has
the vital significance to the practical engineering practice.
For nonlinear continuous-time systems with sampled mea-
surements, many satisfying results have been achieved by
applying different techniques: Kalman filtering technique
[6], discrete-time observer based on exact or approximate
discrete-time description of the continuous dynamics [7]–
[8], continuous-discrete observer design [9]–[10].
Recently, two new approaches have been proposed for the
design of full-order observer of Lipschitz nonlinear systems
with nonuniformly sampled measurements: (i) the observer
is designed by updating the observer state in an impulsive
system fashion [11]–[12]; (ii) the observer is designed by up-
dating the observer state in a sample-and-hold fashion [13].
The main advantage of the two design approaches is that the
proposed observers have simple structures and can be eas-
ily implemented. In [13], the condition for the existence of
sampled-data observer is based on the discontinuous Lya-
punov functional based method introduced in [14]. However,
since the sampled-data control systems are continuous con-
This work was supported in part by the National Natural Science Foun-
dation of China under Grant 61164016 and Guangxi Natural Science Foun-
dation (2011GXNSFA018141 & 2013GXNSFDA019003).
trol systems, the discontinuous Lyapunov functional based
method may neglect some useful information concerning the
dynamical nature of the error dynamics. Therefore, there is
room for reducing the conservatism of the result derived in
[13].
Motivated by the above discussion, we revisit the prob-
lem of observer design for Lipschitz nonlinear systems with
nonuniformly sampled measurements. By introducing a
time-varying Lyapunov functional to analyze the stability of
the observer error dynamics, a new less conservative suffi-
cient condition for the existence of observers with sample-
and-hold is derived. The new condition is dependent on the
upper bound of the sampled intervals and is expressed in
the form of linear matrix inequalities (LMIs). By solving
a set of LMIs, the observer gain matrix can easily be ob-
tained. The effectiveness of the proposed method has been
illustrated through the flexible joint robotic arm and Chua’s
circuit.
Notation The notation M>(≥) 0 is used to denote a
symmetric positive-definite (positive-semidefinite) matrix. I
denotes an identity matrix of appropriate dimension. N de-
notes the set of positive integers. ·denotes the Euclidean
norm for vectors or the spectral norm for matrices.
2 Problem formulation
Consider the following nonlinear continuous-time system
⎧
⎨
⎩
˙x(t)=Ax(t)+ρ(t, u(t)) + Gσ(Hx(t))
y(t)=Cx(t),t= t
k
x(t
0
)=x
0
(1)
where x(t) ∈ R
n
is the state vector, y(t) ∈ R
m
is the mea-
sured output available at discrete time instants, u(t) ∈ R
p
is the control input, A ∈ R
n×n
, G ∈ R
n×m
, H ∈ R
m×n
,
C ∈ R
q×n
are constant matrices, and ρ : R × R
p
−→ R
n
is locally Lipschitz function which depends on the known
arguments. The nonlinear function σ : R
m
−→ R
m
satis-
fies the Lipschitz condition: there exists a positive constant
γ>0 such that
σ(v) − σ(ω)≤γv − ω, ∀v, ω ∈ R
m
. (2)
3URFHHGLQJVRIWKHQG&KLQHVH&RQWURO&RQIHUHQFH
-XO\;LDQ&KLQD
6687