李 賓, 龍述君
(1. 西華大學 理學院, 四川 成都 610039; 2. 樂山師范學院 數學與信息科學學院, 四川 樂山 614004)
一類含有分布時滯的非自治細胞神經網絡的全局指數穩定性
李 賓1, 龍述君2*
(1. 西華大學 理學院, 四川 成都 610039; 2. 樂山師范學院 數學與信息科學學院, 四川 樂山 614004)
研究一類具有分布時滯的非自治細胞神經網絡的全局指數穩定性,通過建立一個新的微分-積分不等式,并將其運用到非自治細胞神經網絡的穩定性研究中,從而得到全局指數穩定性的充分判別準則.結論較以往的文獻結果,更具一般性,適用范圍更廣.最后,通過實例說明所獲結論的可行性和優越性.
細胞神經網絡; 非自治; 全局指數穩定; 微分-積分不等式; 分布時滯.
自從L.O.Chua等1988年在文獻[1-2]中首次提出細胞神經網絡理論以來,其相關理論已經成功應用在模式識別、移動圖像處理、運動目標識別等領域.網絡系統的穩定性是這些應用必須考慮的因素,因此,吸引了很多學者對該動力行為進行研究并獲得許多有用的結果[3-13].在現實生活中,時滯現象是普遍存在的,時滯的存在往往會對系統具有一定的破壞作用,如在神經元間信號傳輸過程中產生的時滯現象會對系統造成震蕩、混沌或不穩定現象,因而研究時滯對動力行為的影響是非常有必要的[14-17].
當考慮一個長期的動力行為時,系統的參數常常受到環境干擾而隨時間變化,非自治微分系統更能準確描述此類情況.對人工細胞神經網絡而言,也不例外.然而對非自治神經網絡的研究遠比對自治神經網絡的研究困難得多.人們嘗試各種方法對其進行研究,并獲得較好結果.文獻[18-19]采用李亞普洛夫楔函數方法研究具有有限時滯的非自治細胞神經網絡的動力行為,獲得了系統穩定性的充分判據;文獻[8-9]通過建立新的微分不等式,運用不等式分析技巧研究幾類非自治神經網絡的動力行為,拓寬了判定動力行為條件的使用范圍.此外,由于具有各種軸突大小和軸突長度的大量平行路徑存在,神經網絡通常具有空間延展,此類情況下,時滯往往以分布時滯呈現出來,因而有必要研究具有分布時滯的非自治細胞神經網絡的動力行為[20-21],但這些成果中都要求判定條件在時間變化范圍內一致成立,這在一定程度上影響了成果的適用范圍.
基于以上分析,本文將對含有分布時滯的非自治細胞神經網絡的全局指數穩定進行研究,得到全局指數穩定性的充分判別準則,從而推廣一些現有文獻的相關結果.
考慮如下含分布時滯的細胞神經網絡模型:


i=1,2,…,n,
(1)


C[X,Y]表示從拓撲空間X到拓撲空間Y的所有連續映射全體.特別地,令CC[(-∞,0],Rn]表示所有有界連續函數φ:(-∞,0]→Rn且|.
定義1如果存在常數λgt;0,M≥1,使得對于系統(1)的任意2個分別滿足初值條件φ,φ∈C的解x(t,φ),y(t,φ).對任意t≥t0有
‖x(t,φ)-y(t,φ)‖≤M‖φ-φ‖e-λ(t-t0),
則稱系統(1)是全局指數穩定的.
引理1假設p(t)滿足如下含有脈沖項的微積分不等式
(2)

如果當t≤t0時有
p(t)≤me-λ(t-t0),
(3)
則
t≥t0,
(4)
其中,m為正常數,δkmax{1,|pk|+|qk|×k(s)eλsds},λ∈(0,λ0),滿足

(5)


下面將證明

(6)
為了證明(6)式,首先證明,對任意的常數εgt;0有

(7)
假設(7)式不成立,則存在一個t*∈(t0,t1),使得
p(t*)=n(t*),D+p(t*)≥n′(t*),
(8)
p(t)≤n(t), t∈(-∞,t*].
(9)
結合(2)、(5)、(7)~(9)式,可以得到

(10)
顯然與(8)式的第二個不等式矛盾,因而(7)式成立,在(7)式中,令ε→0,得到(6)式成立.
接下來,結合(2)、(3)和(6)式,得到

(11)
則

運用與(6)式類似的方法,得到

通過歸納得到,對任意的k∈N有

故原命題得證.
注1在引理1中,當t≥t0時,如果α(t)≡0和β(t)≡0,得到文獻[13]中的引理1.
定理1假設如下條件成立:
(A1) 對任意i,j=1,2,…,n和l=1,2,…,m,存在kj和uijl,使得
|fj(x)-fj(y)|≤kj|x-y|,
|gijl(x)-gijl(y)|≤uijl|x-y|;


(12)

(13)
(A3) 存在常數λgt;γ≥0和h≥0,使得
(14)
其中λ滿足
(15)
則系統(1)是全局指數穩定的且指數收斂率不低于λ-γ.
證明設x(t),y(t)是系統(1)的任意2個分別滿足初值條件φ,φ∈C的解.令

結合系統(1)和條件(A1)、(A2),得到

(16)
因為φ,φ∈C,則存在一個正數M≥1,使得





(17)

考慮如下二維含有分布時滯的非自治細胞神經網絡系統
(18)

明顯地,gi(s)滿足李普希茲條件且ui=1(i=1,2),得到(A2)的參數
-b1(t)≤-2.25+δ(t),
-b2(t)≤-2.5+δ(t);
|c11(t)|+|c21(t)|≤1.75,
|c12(t)|+|c22(t)|≤1.6.


下面計算

對任意tgt;t0≥0,存在正整數n≥m≥0,使得nT≤tlt;(n+1)T,mT≤t0lt;(m+1)T;令t=nT+u,t0=mT+w,其中0≤u,wlt;T.通過計算得到


特別地,當k∈N時,令
明顯地,δ(s)的周期為2π,易得



圖 1 x1(t)的狀態曲線

圖 2 x2(t)的狀態曲線

圖 3 ‖x(t)-y(t)‖的衰減曲線
注3在文獻[20]的條件(H2)中,令ωi=qij=rij=1(i,j=1,2);得到相應的判別條件:
h1(t)=b1(t)-(|c11(t)|+|c21(t)|)=
1.625-cost-sint-δ(t);
h2(t)=b2(t)-(|c12(t)|+|c22(t)|)=
1.75-0.75sint-δ(t).
通過觀察圖4和5發現,對任意的t≥t0,不存在σgt;0,使得h1(t)gt;σ,h2(t)gt;σ成立,因此,文獻[20]中的結論對此例是失效的.

圖 4 h1(t)對應的圖形

圖 5 h2(t)對應的圖形
本文研究一類含有分布時滯的非自治細胞神經網絡的全局指數穩定性問題,通過運用不等式分析技巧,建立一個新的微分-積分不等式,使得神經網絡的全局指數穩定的判別準則得到了進一步放松,較之前的結果適用范圍更廣.
[1] CHUA L O, YANG L. Cellular neural networks:theory.IEEE trans[J]. Circuits Syst,1988,35(I):1257-1272.
[2] CHUA L O, YANG L. Cellular neural networks:applications.IEEE trans[J]. Circuits Syst,1988,35(I):1273-1290.
[3] ROSKA T, BOROS T, THIRAN P, et al. Detecting simple motion using cellular neural networks[C]//IEEE International Workshop on Cellular Neural Networks and Their Applications,1990. Cnna-90 Proceedings. IEEE,1990:127-138.
[4] CHUA L O, ROSKA T. Cellular neural networks with nonlinear and delay-type template elements[C]//EE Int Workshop Cellular Neural Networks Applications, Cnna-90 Proceeding,1991:12-25.
[5] LIANG J L, CAO J D. Global asymptotic stability of bi-directional associative memory networks with distributed delays[J]. Appl Math Comput,2004,152(1):415-424.
[6] SONG Q K, YAN H, ZHAO Z J, et al. Global exponential stability of complex-valued neural networks with both time-varying delays and impulsive effects[J]. Neural Networks,2016,79(1):108-116.
[7] JIANG M H, MU J, HUANG D S. Globally exponential stability and dissipativity for nonautonomous neural networks with mised time-varying delays[J]. Neurocomputing,2016,205(1):421-429.
[8] LONG S J, XU D Y. Global exponential stability of non-autonomous cellular neural networks with impulses and time-varying delays[J]. Commun Nonlinear Sci Numer Simulat,2013,18(1):1463-1472.
[9] LONG S J, LI H H, ZHANG Y X. Dynamic behavior of nonautonomous cellular neural networks with time-varying delays[J]. Neurocomputing,2015,168(1):846-852.
[10] CHEN Z W, YANG J, ZHONG S M. Delay-partitioning approach to stability analysis of generalized neural networks with time-varying delay via new integral inequality[J]. Neurocomputing,2016,191(1):380-387.
[11] LONG S J, XU D Y, ZHU W. Global exponential stability of impulsive dynamical systems with distributed delays[J]. Electronic Qualitative Theory of Differential Equations,2007,10(1):1-13.
[12] 龍述君,向麗. 一類具有分布時滯的Hopfleld神經網絡的穩定性[J]. 四川師范大學學報(自然科學版),2006,29(5):566-569.
[13] XU D Y, ZHU W, LONG S J. Global exponential stability of impulsive integro-differential equation[J]. Nonlinear Analysis,2006,64(1):2805-2816.
[14] LOU X Y, CUI B T. Boundedness and exponential stability for nonautonomous RCNNs with distributeddelays[J]. Comput Math Appl,2007,54(1):589-598.
[15] LI L L, JIAN J G. Exponential p-convergence analysis for stochastic BAM neural networks with time-varying and infinite distributed delays[J]. Appl Math Comput,2015,266(1):860-873.
[16] 楊德剛. 一種新的時滯細胞神經網絡全局漸進穩定性準則[J]. 重慶師范大學學報(自然科學版),2007,24(3):46-50.
[17] 李建軍,楊志春. 雙向聯想記憶神經網絡的指數輸入-狀態穩定性[J]. 重慶師范大學學報(自然科學版),2016,33(4):79-84.
[18] LI L L, JIAN J G. Delay-dependent passivity analysis of impulsive neural networks with time-varying delays[J]. Neurocomputing,2015,168(1):276-282.
[19] CHEN G L, XIA J W, ZHUANG H M. Improved passivity analysis for neural networks with Markovian jumping parameters and interval time-varyingdelays[J]. Neurocomputing,2015,155(1):253-260.
[20] JIANG H J, TENG Z D. Boundeness and global stability for nonautonomous recurrent neural networks with distributed delays[J]. Chaos, Solitons and Fractals,2006,30(1):83-93.
[21] ZHANG Q, WEI X P, XU J. Global exponential stability for nonautonomous cellular neural networks with unbounded delays[J]. Chaos, Solitons and Fractals,2009,39(1):1144-1151.
[22] LI X D. Existence and global exponential stability of periodic solution for impulsive Cohen-Grossberg-type BAM neural networks with continuously distributed delays[J].Appl Math Comput,2009,215(1):292-307.
[23] HUANG C X, CAO J D. Almost sure exponential stability of stochastic cellular neural networks with unbounded distributed delays[J]. Neurocomputing,2009,72(1):3352-3356.
[24] RAKKIYAPPAN R, BALASUBRAMANIAM P. LMI conditions for stability of stochastic recurrent neural networks with distributed delays[J]. Chaos Solitons and Fractals,2009,40(1):1688-1696.
[25] ESTEVES S. Global exponential stability of non-autonomous neural network models with continuous distributed delays[J]. Appl Math Comput,2013,219(17):9296-9307.
[26] OLIVEIRA JOSE J. Global stability of a Cohen-Grossberg neural network with both time-varying and continuous distributed delays[J]. Nonlinear Analysis:Real World Applications,2011,12(1):2861-2870.
[27] MOHAMAD SANNAY. Global exponential stability in DCNNs with distributed delays and unbounded activation[J]. J Comput Appl Math,2007,205(1):161-173.
[28] OLIVEIRA JOSE J. Global asymptotic stability for neural network models with distributed delays[J]. Math Comput Modeling,2009,50(1):81-91.
[29] LI Y T, WANG J Y. An analysis on the global exponential stability and the existence of periodic solutions for non-autonomous hybrid BAM neural networks with distributed delays and impulses[J]. Comput Math Appl,2008,56(1):2256-2267.
2010MSC:34D23; 92B20
(編輯 鄭月蓉)
Global Exponential Stability of Non-autonomous Cellular Neutral Network Models with Distributed Delays
LI Bin1, LONG Shujun2
(1.SchoolofScience,XihuaUniversity,Chengdu610039,Sichuan;2.CollegeofMathematicsandInformationScience,LeshanNormalUnivesity,Leshan614004,Sichuan)
In this paper, we investigate the global exponential stability of non-autonomous cellular neural networks with distributed delays. We establish a new differential-integro inequality and use it in the investigation the stability of cellular neural networks to obtain a sufficient condition for the global exponential stability for the considered system. Our results improve the known results in the literature. Finally, an example is given to illustrate the effectiveness and superiority of our conclusion
cellular neural network; non-autonomous; global exponential stability; differential-integro inequality; distributed delays
O175.13
A
1001-8395(2017)06-0780-07
10.3969/j.issn.1001-8395.2017.06.012
2016-10-10
四川省教育廳創新團隊項目(16TD0029)
*通信作者簡介:龍述君(1975—),男,教授,主要從事運籌學與控制論的研究,E-mail:longer207@yahoo.com.cn