Brief Papers
Fuzzy observer-based output feedback control design of discrete-time
nonlinear systems: An extended dimension approach
Qingzhong Gao
a,b,
n
, Dazhi Wang
a
, Yan Zhao
b
a
College of Information Science and Engineering, Northeastern University, Shenyang 110004, China
b
Academy of Automation, Shenyang Institute of Engineering, Shenyang 110136, China
article info
Article history:
Received 11 August 2015
Received in revised form
28 February 2016
Accepted 1 March 2016
Available online 16 March 2016
Keywords:
Fuzzy logic
Nonlinear systems
Asymptotic stability
Observer-based control
abstract
Recently, previous results of fuzzy observer-based output feedback control design of discrete-time
nonlinear systems have been relaxed by using a multi-samples’ method but it is not very mature. This
study presents some developments which give a family of feasible solutions to improve the design
quality of the recent result. To accomplish this work, an systematic multi-samples’ approach that is
parameter-dependent on normalized fuzzy weighting functions on optional multi-samples’ points is
presented in the interest of making good use of more helpful information of the discrete-time nonlinear
systems. Furthermore, a new Lyapunov function candidate is given to cooperate with the proposed
approach while those redundant terms composed of a set of combinations of the tþ1 sampled point are
removed. As a result of the above efforts, the main defects of the recent result can be overcome and its
conservatism is further reduced by an efficient way. In particular, a compromising solution in aspect of
not only reducing the conservatism but also suppressing the computational burden is obtained in the
special case of m¼1. Finally, an illustrative example is given to validate the effectiveness of the approach
presented in this study.
& 2016 Elsevier B.V. All rights reserved.
1. Introduction
Fuzzy logic that is made of fuzzy rules based on linguistic
knowledge is one complement to mathematical knowledge in
addressing systems whose dynamics are not so well understood
and whose models cannot be so conveniently established [1,2].In
1985, the Takagi–Sugeno (T–S) fuzzy model given in [3] was pro-
posed to cope with the analysis of nonlinear dynamical systems.
Today's nonlinear control studies [4–10] are highly based on the
T–S fuzzy model. Owing to the usage of the so-called single
quadratic Lyapunov function, finding Lyapunov functions is still a
hard mission, particularly, for complicated T–S fuzzy systems [11].
A main reason for the difficulty is that the requirement of a
monotonically decreasing condition with respect to time for a
given single Lyapunov matrix is sometimes too strict [12]. For the
sake of relaxing this restriction, different kinds of nonparallel
distributed compensation control laws have been proposed in
recent years, see [13–17] and the cited literature therein. Because
more information of the current-time normalized fuzzy weighting
function is involved into the process of control design, the
conservatism of control synthesis of T–S fuzzy control systems has
been becoming more and more relaxed since the past five years.
Usually, most of the aforementioned research literature are on
the basis of the assumption that all the system states are mea-
surable, which is not satisfied for most practical systems. In order
to overcome this obstacle, a lot of related research works, for
instance, filter designs based on the T–S fuzzy model [18–23],
different fuzzy observer designs [24–27], static output feedback
control of T–S fuzzy systems [28], dynamic output feedback con-
trol of T–S fuzzy systems [29], fuzzy observer-based output feed-
back control designs [30,31], have been developed in recent years.
Particularly, the obtained design conditions in [30] were presented
for calculating a family of observer-based controllers by means of
bilinear matrix inequalities which were not convex and NP-hard to
be implemented. As an alternative, two-step procedure was often
used but much conservatism was inevitably introduced. Recently,
a single-step linear matrix inequality method has been presented
in [31]. Using the result given in [31]
, both the controller and
observer parameters have been provided by means of a family of
strict linear matrix inequalities, which are numerically feasible
with the help of commercially available software. It is noteworthy
that the recent approach proposed in [31] is not very mature as it
can be summarized by two main reasons as follows: (A) Despite
announcing that a multi-samples’ method is applied, there are
only three sampled points involved in the result of [31], i.e.,
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/neucom
Neurocomputing
http://dx.doi.org/10.1016/j.neucom.2016.03.004
0925-2312/& 2016 Elsevier B.V. All rights reserved.
n
Corresponding author at: College of Information Science and Engineering,
Northeastern University, Shenyang 110004, China.
E-mail address: qingzhonggaoneu@163.com (Q. Gao).
Neurocomputing 196 (2016) 167–174