Elsevier

Fuzzy Sets and Systems

Volume 394, 1 September 2020, Pages 40-64
Fuzzy Sets and Systems

Non-fragile memory filtering of T-S fuzzy delayed neural networks based on switched fuzzy sampled-data control

https://doi.org/10.1016/j.fss.2019.09.001Get rights and content

Abstract

This paper deals with the non-fragile memory filtering issue of T-S fuzzy delayed neural networks with randomly occurring time-varying parameters uncertainties and variable sampling rates. Compared with existing sampled-data control schemes, an improved switched fuzzy memory sampled-data control protocol is designed for the first time, which involves not only a signal transmission delay but also switched topologies. By developing some new terms and taking full advantage of the variable characteristics related to the actual sampling pattern, a modified loose-looped fuzzy membership functions (FMFs) dependent Lyapunov-Krasovskii functional (LKF) is constructed based on the information of the time derivative of FMFs. Meanwhile, some relaxed matrices chosen in LKF are not consequentially positive definite. Moreover, with the LKF methodology and employing the developed estimation technique, several optimized control algorithms with both a larger sampling period and upper bound of time-varying delays for achieving the stabilization of the resultant T-S fuzzy delayed neural networks are derived. Finally, a numerical example is presented to demonstrate the superiority and applicability of the theoretical results.

Introduction

Over the last there decades, neural networks (NNs) have garnered wide-scale attention since they have been comprehensively investigated and developed in a variety of science and engineering fields, such as signal processing, pattern recognition, smart antenna arrays, image processing, associative memory, industrial automation, combinatorial optimization problems, and so on [1], [2], [3], [4]. Meanwhile, it is a well known phenomenon that time delay necessarily exists in natural and man-made systems and cannot be neglected, for example, time delay effects can not be omitted in the communication systems mainly due to the limited switching speed of the hardware [5]. However, the existence of time delay often results in undesired dynamics, including divergence, oscillation, or even instability of NNs [6], [7], [8], [9], [10]. Hence, it is all-important to study the stability of delayed neural networks (DNNs). This issue has recently attracted increasing attention, a great number of literatures about the dynamic behaviors of DNNs have been published and have left many important results [11], [12], [13], [14], [15], [16].

In the mathematical modeling of DNNs, parameter variations are also ubiquitous due to the system structure variation, environmental disturbance, parameter fluctuations and estimation errors, which would result in complex dynamical behaviors [17], [18], [19]. Therefore, we must take full account of the parameter uncertainties, when considering stability of DNNs, in which case, one must deal with the robust stability of DNNs. In recent years, many robust stability algorithms and criteria have been derived in [20], [21], [22], [23].

Meanwhile, fuzzy systems in the form of the Takagi-Sugeno fuzzy model (TSFM) [24] have provoked rapidly widespread interests from mathematics and engineering fields in recent years [25], [26]. Because T-S fuzzy systems (TSFSs) have excellent ability in analyzing, synthesizing, and approximating complex dynamical behaviors by a set of IF-THEN rules [27], [28]. Consequently, finite-time stability analysis, fault-tolerant control, reliable dissipative filtering, and integral sliding mode control, memory feedback control problems for TSFSs have gained increasing attention, and many effective methods and important results have been reported in [29], [30], [31], [32], [33], [34]. Based on the above discussion, the concept of incorporating fuzzy logic has been extended to describe DNNs via TSFM and some stability criteria have been derived [35], [36], [37], [38], [39], [40], [41]. Whereas it is worth noting that all the methods mentioned above on the stability analysis of TSFDNNs are independent of fuzzy membership functions (FMFs). Besides, some related processing techniques are scarce, such as the Wirtinger's inequality and time-dependent Lyapunov-like functional [35], [36], [37], [38], [39], [40], [41]. To this end, a good issue is whether we can propose an FMFs dependent approach by taking these neglected factors into account to develop less conservative conditions for TSFDNNs? This is the first motivation for this paper.

On the other hand, with the development of communication and digital technology, sampled-data control (SDC) theory has attracted much attention [42], [43], [44], [45], [46]. Compared with some continuous control [35], [36], [37], [38], [39], [40], [41], the SDC shows more advantages, such as efficiency, high reliability, and easy installation. Accordingly, the investigation of fuzzy SDC of TSFDNNs is very important. Very recently, many interesting stability conditions have been obtained on the filtering issue of TSFDNNs via fuzzy SDC [47], [48], [49]. By using LKF together with the zero function, the authors considered the sampled-data filtering of uncertain TSFNNs with interval time-varying delays [47]. Based on employing the Newton-Leibniz formulation and SDC way, the authors obtained a good design of a state estimator for TSFNNs with Markovian jumping parameters [48]. It should be noted that the relationships of e(t), e(tτ(t)) and 1τ(t)tτ(t)te(s)ds were not taken fully considered [47], [48]. In order to reduce the conservatism of obtained results [47], [48], the authors investigated the sampled-data filtering problem of TSFDNNs by using the relaxed reciprocally convex inequality and some novel integral inequalities [49]. Nevertheless, these approaches [47], [48], [49] suffer from the following drawbacks: (1) the chosen LKFs in [47], [48], [49] needs to satisfy the positive condition inside the sampling intervals; (2) the information on the actual sampling pattern has not been fully utilized in [47], [48], [49], which may result in some conservatism; (3) the fuzzy SDC with a constant signal transmission delay in [50] have not yet considered. Thus, how to construct the appropriate LKF for the sampled-data filtering problem of TSFNNs to get a larger sampling period is the another motivation of this paper.

Motivated by the above consideration, in this paper, our attention focuses mainly on investigating the switched fuzzy non-fragile SDC issue for TSFDNNs with randomly occurring time-varying parameters uncertainties (ROTVPUs) and variable sampling rates. A new switched fuzzy memory SDC scheme that is more general than the conventional SDC one is used to deal with this problem. By constructing a augmented loose-looped FMFs dependent LKF that involves some novel terms and relaxed matrices, several optimized control algorithms are presented such that the resultant TSFDNNs is stable. The primary competitive advantage of the modified loose-looped FMFs dependent LKF is that it enables us to take full advantage of the variable characteristics related to the actual sampling pattern and the information of the time derivative of FMFs. The main contributions of this paper are as follows: (1) A switched fuzzy memory SDC scheme is firstly developed for the resultant TSFDNNs; (2) A new loose-looped FMFs dependent LKF with relaxed conditions is constructed; (3) Several optimized control algorithms with both a larger sampling period and upper bound of time-varying delays are obtained for the resultant TSFDNNs.

Notation: Notations used in this paper are fairly standard: Rn denotes the n-dimensional Euclidean space, Rn×m is the set of all n×m dimensional matrices; I denotes the identity matrix of appropriate dimensions, T stands for matrix transposition, the natation X>0 (respectively X0), for XRn×n means that the matrix is real symmetric positive definite (respectively, positive semi-definite); diag{r1,r2,,rn} denotes block diagonal matrix with diagonal elements ri,i=1,2,,n, the symbol ⁎ represents the elements below the main diagonal of a symmetric matrix, Sym{M} is defined as Sym{M}=12(M+MT).

Section snippets

Preliminaries

Consider the following TSFDNNs with ROTVPUs:{y˙(t)=(C+α1(t)C(t))y(t)+(A+α2(t)A(t))f(y(t))+(B+α3(t)B(t))f(y(tτ(t)))+u(t)+J,z(t)=Hy(t),y(t)=φ(t),t[τ2,0], where y(t)Rn is the neuron state vector of the network, z(t)Rm is the measurement output, f(y(t))=[f1(y1(t)),f2(y2(t)),,fn(xn(t)]TRn is the neuron activation function, φ(t)Rn is the in initial state vector. f(y(tτ(t)))=[f1(y1(tτ(t))),f2(y2(tτ(t))),,fn(yn(tτ(t))]TRn, J=[J1,J2,,Jn]T is a constant input vector, u(t) is the

Main results

In this section, we will give the stability analysis of uncertain TSFDNNs filter. We first give a better estimation algorithm design for the close-loop TSFDNNs filter to be asymptotically stable. When Ci(t)=0, Ai(t)=0, Bi(t)=0, K1(t)=0, K2(t)=0, απ(t)=0 and βσ(t)=0, the resultant TSFDNNs (22) can be rewritten ase˙(t)=i=1rhi(ω(t))j=1rhj(ω(tk))[Cie(t)+Aig(e(t))+Big(e(tτ(t)))+K1jHie(td(t))+K2jHie(td(t)δ)]. For the sake of simplicity of matrix representation, eiT=[0,,I,,0] (i=1,,19)

Numerical examples

In this section, one numerical example is presented to demonstrate the potential benefits and effectiveness of our main results for TSFDNNs.

Example 1

Consider the following fuzzy non-fragile memory SDC filtering of TSFDNNs with switched topologies (r=2):

IF ω1(t) is Mi1 and ⋯ωp(t) is Mip THEN

Fuzzy  Rule 1:

IF ω1(t) is M11 and ⋯ωp(t) is M1p THEN{y˙(t)=C1y(t)+A1f(y(t))+B1f(y(tτ(t)))+u(t),z(t)=H1y(t),

Fuzzy  Rule 2:

IF ω2(t) is M21 and ⋯ ωp(t) is M2p THEN{y˙(t)=C2x(t)+A2f(y(t))+B2f(y(tτ(t)))+u(t),z(t)=H2y(t)

Conclusions

In this paper, we have investigated the non-fragile memory filtering problem of TSFDNNs with ROTVPUs and variable sampling rates. Under the consideration of communication delay and state switched signals, we proposed a new fuzzy switched memory sampled-data control scheme with variable sampling. Together with FMFs dependent LKF and some novel integral inequality, we designed improved control algorithms. The relaxed conditions derived in this paper make full use of the variable characteristics

References (72)

  • K.B. Shi et al.

    New reliable nonuniform sampling control for uncertain chaotic neural networks under Markov switching topologies

    Appl. Math. Comput.

    (2019)
  • Z.P. Wang et al.

    Fuzzy impulsive control for uncertain nonlinear systems with guaranteed cost

    Fuzzy Sets Syst.

    (2016)
  • J. Cheng et al.

    Finite-time H fuzzy control of nonlinear Markovian jump delayed systems with partly uncertain transition descriptions

    Fuzzy Sets Syst.

    (2017)
  • P. Balasubramaniam et al.

    Delay-dependent robust exponential state estimation of Markovian jumping fuzzy Hopfield neural networks with mixed random time-varying delays

    Commun. Nonlinear Sci. Numer. Simul.

    (2011)
  • S.J. Long et al.

    Global exponential p-stability of stochastic non-autonomous Takagi-Sugeno fuzzy cellular neural networks with time-varying delays and impulses

    Fuzzy Sets Syst.

    (2014)
  • R. Samidurai et al.

    Effects of leakage delays and impulsive control in dissipativity analysis of Takagi-Sugeno fuzzy neural networks with randomly occurring uncertainties

    J. Franklin Inst.

    (2017)
  • C.K. Ahn

    Passive and exponential filter design for fuzzy neural networks

    Inf. Sci.

    (2013)
  • T.H. Lee et al.

    Improved criteria for sampled-data synchronization of chaotic Lur'e systems using two new approaches

    Nonlinear Anal. Hybrid Syst.

    (2017)
  • R.M. Zhang et al.

    A novel approach to stability and stabilization of fuzzy sampled-data Markovian chaotic systems

    Fuzzy Sets Syst.

    (2018)
  • R.M. Zhang et al.

    New approach on designing stochastic sampled-data controller for exponential synchronization of chaotic Lur'e systems

    Nonlinear Anal. Hybrid Syst.

    (2018)
  • E. Yucel et al.

    Sampled-data filtering of Takagi-Sugeno fuzzy neural networks with interval time-varying delays

    Fuzzy Sets Syst.

    (2017)
  • Y.J. Liu et al.

    Nonfragile exponential synchronization of delayed complex dynamical networks with memory sampled-data control

    IEEE Trans. Neural Netw. Learn. Syst.

    (2018)
  • K. Liu et al.

    Wirtinger's inequality and Lyapunov-based sampled-data stabilization

    Automatica

    (2012)
  • A. Seuret et al.

    Wirtinger-based integral inequality: application to time-delay systems

    Automatica

    (2013)
  • H.B. Zeng et al.

    A new looped-functional for stability analysis of sampled-data systems

    Automatica

    (2017)
  • J. Cheng et al.

    Quantized H filtering for switched linear parameter-varying systems with sojourn probabilities and unreliable communication channels

    Inf. Sci.

    (2018)
  • J.Y. Xiao et al.

    Extended dissipative conditions for memristive neural networks with multiple time delays

    Appl. Math. Comput.

    (2018)
  • H.B. Zeng et al.

    Sampled-data-based dissipative control of T-S fuzzy systems

    Appl. Math. Model.

    (2019)
  • K. Mathiyalagan et al.

    Non-fragile H synchronization of memristor-based neural networks using passivity theory

    Neural Netw.

    (2016)
  • J. Xiao et al.

    Extended dissipative state estimation for memristive neural networks with time-varying delay

    ISA Trans.

    (2016)
  • P. Selvaraj et al.

    Finite-time synchronization of stochastic coupled neural networks subject to Markovian switching and input saturation

    Neural Netw.

    (2018)
  • T.P. Zhang et al.

    Topology preserving non-negative matrix factorization for face recognition

    IEEE Trans. Image Process.

    (2008)
  • S. Arik

    New criteria for global robust stability of delayed neural networks with norm-bounded uncertainties

    IEEE Trans. Neural Netw. Learn. Syst.

    (2014)
  • Z.G. Wu et al.

    Local synchronization of chaotic neural networks with sampled-data and saturating actuators

    IEEE Trans. Cybern.

    (2014)
  • T.H. Lee et al.

    Extended dissipative analysis for neural networks with time-varying delays

    IEEE Trans. Neural Netw. Learn. Syst.

    (2014)
  • X.Z. Liu et al.

    Pinning impulsive synchronization of reaction–diffusion neural networks with time-varying delays

    IEEE Trans. Neural Netw. Learn. Syst.

    (2017)
  • Cited by (238)

    View all citing articles on Scopus

    This work was supported by the National Natural Science Foundation of China under Grant (No. 61703060, 61802036, 61701048, 11601474 and 11461082, 61873305), and the Fundamental Research Funds for the Central Universities, Southwest Minzu University (2019NQN07), the Opening Fund of Geomathematics Key Laboratory of Sichuan Province (scsxdz2018zd02 and scsxdz2018zd04), Guangxi Natural Science Foundation Project (No. 2017GXNSFBA198179) and Basic Ability Promotion Project for Young and Middle-aged Teachers in Universities of Guangxi (No. 2018KY0214).

    View full text