New Results on
filter Design for Nonlinear Time-Delay
Systems Via Fuzzy Line-Integral Approach
Ziye Zhang
1,2
•
Chong Lin
2
•
Bing Chen
2
Received: 19 August 2015 / Revised: 19 October 2015 / Accepted: 10 December 2015 / Published online: 12 January 2016
Ó Taiwan Fuzzy Systems Association and Springer-Verlag Berlin Heidelberg 2016
Abstract This paper revisits the problem of H
1
filtering
design for nonlinear systems with time-varying delay
through Takagi–Sugeno (T–S) fuzzy models. By applying
the fuzzy line-integral Lyapunov function, a new sufficient
condition for the existence of the fuzzy H
1
filter design is
established in terms of linear matrix ineq ualities. The
present method provides improvements and produces better
results than existing ones in the literature. Two examples
are given to show the advantages and effectiveness of the
proposed results.
Keywords H
?
filter Nonlinear systems Time delay
Fuzzy line-integral approach
1 Introduction
The filtering problem is an important issue with many
practical applications in signal processing and it has
attracted considerable interest through various methods,
such as the Kalman filter approach [1, 2] and the H
1
filter
approach [3–18]. In contrast to the Kalman filter, the main
advantage of the H
1
filter is that it can determine an
asymptotically stable filter with uncertain noise signal
statistics. Moreover, the well-known Takagi–Sugeno (T–S)
fuzzy mode l approach [19] has been recognized as an
effective tool in approximating nonlinear systems by linear
subsystems [20], and such type of systems can be dealt
with theories extended from linear systems [21]. Hence,
various methods have been developed for the analysis and
synthesis of T–S fuzzy systems [2, 7–36]. Among these
developments, as time delay often appears in practical
systems and it is a major cause of ins, the H
1
filter design
for T–S fuzzy systems with time delay has been a hot topic
in the past few years, see [7–18].
For the stability analysis and control synthesis of T–S
fuzzy systems, a fuzzy weighting-dependent Lyapunov
function is proposed in [23], which is effective in reducing
the conservatism. However, this technique is based on the
assumption that the upper bound for the time derivatives of
membership functions is known [24–27]. In fact, the
assumption is suitable for a narrow class of systems. To
remove this assumption, a fuzzy line-integral Lyapunov
function is presented in [28] and this idea has been used in
[29–32] for stability analysis and fuzzy control design.
Meanwhile, for the corresponding H
1
filtering of delayed
fuzzy systems, the quadratic Lyapunov–Krasovskii func-
tional (LKF) approach is popular and it results in various
approaches [8–11 , 13–16]. It is known that these results are
comparably conservative as common Lyapunov matrices
are chosen. In addition, the fuzzy weighting-dependent
Lyapunov functionals are adopted for designi ng the H
1
filter in [7, 10, 17]. A new fuzzy LKF is chosen in [12] and
it leads to certain improvement for the H
1
filter design.
However, many decision variables are needed therein and
much computational burden is involved because of the
employment of the convex method and the fuzzy free-
& Chong Lin
linchong_2004@hotmail.com
Ziye Zhang
zhangzy02@126.com
Bing Chen
chenbing1958@126.com
1
College of Mathematics and Systems Science, Shandong
University of Science and Technology, Qingdao 266590,
People’s Republic of China
2
Institute of Complexity Science, Qingdao University,
Qingdao 266071, People’s Republic of China
123
Int. J. Fuzzy Syst. (2016) 18(5):904–913
DOI 10.1007/s40815-015-0126-0