Physics Letters A 374 (2010) 4397–4405
Contents lists available at ScienceDirect
Physics Letters A
www.elsevier.com/locate/pla
New delay-dependent stability criteria for neural networks with time-varying
interval delay
✩
Jie Chen
a
, Jian Sun
a,∗
,G.P.Liu
b,c
, D. Rees
b
a
School of Automation, Beijing Institute of Technology, Beijing, 100081, China
b
Faculty of Advanced Technology, University of Glamorgan, Pontypridd CF37 1DL, UK
c
CTGT Center in Harbin Institute of Technology, Harbin, 150001, China
article info abstract
Article history:
Received 21 December 2009
Received in revised form 26 July 2010
Accepted 30 August 2010
Available online 6 September 2010
Communicated by R. Wu
Keywords:
Delay-dependent stability
Time-varying delay
Neural networks
Linear matrix inequality (LMI)
The problem of stability analysis of neural networks with time-varying delay in a given range is
investigated in this Letter. By introducing a new Lyapunov functional which uses the information on the
lower bound of the delay sufficiently and an augmented Lyapunov functional which contains some triple-
integral terms, some improved delay-dependent stability criteria are derived using the free-weighting
matrices method. Numerical examples are presented to illustrate the less conservatism of the obtained
results and the effectiveness of the proposed method.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Neural networks have been applied in many areas such as pattern recognition, data mining, signal filtering, financial prediction and
adaptive control. Since there inevitably exist integration and communication delay, stability of the delayed neural network has been
extensively studied. Existing stability criteria can be classified into two categories, namely, delay-independent ones [1–4] and delay-
dependent ones [5–21]. Since delay-independent stability criteria are usually conservative than delay-dependent ones especially when the
delay is small, delay-dependent stability criteria have received much attention.
By introducing a new Lyapunov functional and using the S-procedure, a less conservative stability condition was put forward in [8].
In order to avoid the conservatism involve by model transformation and bounding techniques for cross terms, free-weighting matrices
method was used to derive stability criteria for neural networks with time-varying delay [9].Resultsin[9] were further improved in
[10] by considering some useful terms which were ignored in previous results when estimating the upper bound on the derivative of
the Lyapunov functional. Using Jensen’s inequality, some simplified stability criteria were proposed [22]. These criteria were equivalent
to those in [10] but with less decision variables. By constructing an augmented Lyapunov functional, improved stability conditions have
been established in [23]. Using the relationship that d
(t) + (h
2
− d(t)) = h
2
and (d(t) − h
1
) + (h
2
− d(t)) = h
2
− h
1
,someimproveddelay-
dependent stability criteria were proposed in [11]. However, the above results are still conservative to some extent and there exists room
for further improvement.
In practice, the lower bound of the delay is not always 0. Therefore, the delay considered in this Letter is assumed to belong to
a given interval. By introducing a new Lyapunov functional, less conservative results are obtained using the free-weighting matrices
method and the idea of convex combination [16]. Using the augmented Lyapunov functional approach, the obtained results are further
improved. Two numerical examples are given to show the effectiveness of the proposed method and the less conservatism of the obtained
results.
✩
This work is partially supported by the Beijing Education Committee Cooperation Building Foundation Project XK100070532 and National Science Foundation for
Distinguished Young Scholars of China under Grant 60925011. The work of G.P. Liu is supported in part by the National Natural Science Foundation of China under Grant
60934006.
*
Corresponding author. Tel.: +86 10 68912464; fax: +86 10 68918232.
E-mail addresses: chenjie@bit.edu.cn (J. Chen), helios1225@yahoo.com.cn (J. Sun), gpliu@glam.ac.uk (G.P. Liu), drees@glam.ac.uk (D. Rees).
0375-9601/$ – see front matter
© 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.physleta.2010.08.070