康 衛(wèi),李林國
(阜陽師范學院信息工程學院,安徽阜陽 236041)
離散分布時滯隨機神經網絡的穩(wěn)定性
康 衛(wèi),李林國
(阜陽師范學院信息工程學院,安徽阜陽 236041)
主要研究了具有分布時滯的隨機系統(tǒng)的全局指數(shù)魯棒穩(wěn)定性,依據(jù)李雅普諾夫方法和線性矩陣不等式的方法,得到了參數(shù)不確定時滯相關的全局均方指數(shù)穩(wěn)定性準則,數(shù)值實例演示證明其結果的有效性和可行性.
神經網絡;離散分布時滯;指數(shù)穩(wěn)定;線性矩陣不等式
近年來,神經網絡理論被廣泛地應用于模式識別、信號處理等領域,國內外許多專家學者和工程技術人員對神經網絡及其應用進行了研究,并得到了一些有用的結論[1].然而,在實際應用過程中,隨機因素總是不可避免地產生,時滯和外界的隨機擾動以及系統(tǒng)參數(shù)的不確定性都是引起系統(tǒng)穩(wěn)定性的重要因素[2-3],文獻[4]分析了一類帶有時滯的隨機神經網系統(tǒng)的全局漸進穩(wěn)定問題,文獻[5]分析了一類隨機神經網絡系統(tǒng)的魯棒穩(wěn)定性問題.然而對于帶有分布時滯的隨機系統(tǒng)的穩(wěn)定性研究并不多,因此筆者運用Lyapunov泛函、線性矩陣不等式以及隨機分析理論分析了全局、魯棒漸近穩(wěn)定性問題,并通過1個實例論證了其可行性.
考慮如下的不確定性的具有分布時滯隨機神經網絡:
其中:x(t)=[x1(t),x2(t),…xn(t)]T∈Rn是狀態(tài)向量;矩陣C=diag {c1,c2,…cn}ci>0;h(t)和τ(t)為變時滯;w(t)=[ω1(t),ω2(t),…,ωn(t)]T是定義在完全概率空間(Ω,F(xiàn),P)的布朗運動.
假設1 時滯h(t)是微分函數(shù),τ(t)為非負有界,且滿足0≤h(t)≤h,h′(t)≤u<1,0≤τ(t)≤τ-.
假設2[5]神經元激勵函數(shù)是有界且滿足其中和(i=1,2,…,n)是常數(shù).
假設3[3]不確定項ΔC(t),ΔA(t),ΔB(t),ΔD(t),ΔH1(t)和ΔH2(t)滿足下面的條件:
其中M,Ni(i=1,2,3,4,5,6)是具有適當維數(shù)的常數(shù)矩陣,且FT(t)F(t)≤I.
引理1 設常數(shù)h>0,向量值函數(shù)f:[0,h]→Rm可積,則對任意正定矩陣R∈Rn×n,
根據(jù)假設(3)式即得.因此系統(tǒng)(1)是魯棒指數(shù)穩(wěn)定的.
[1] WAN L,SUN J.Mean Square Exponential Stability of Stochastic Delayed Hopfield Neural Networks[J].Phys.Lett.A.,2005,343:306-318.
[2] VIMAL SINGH.A Novel Global Robust Stability Criterion for Neural Networks with Delay[J].Physics Letters A,2005,337:369-373.
[3] HUANG H,F(xiàn)ENG G.Delay-Dependent Stability Analysis for Uncertain Stochastic Neural Networks with Time-Varying Delay[J].Physics Letters A,2007,381:93-104.
[4] WANG Z,LAURIA S,LIU X.Exponential Stability of Uncertain Stochastic Neural Networks with Mixed Time-Delays[J].Chaos Solitons Fractals,2007,32(1):62-72.
[5] HUANG H,F(xiàn)ENG G.Delay_Dependent Stability Analysis for Uncertain Stochastic Neural Networks with Time-Varying Delay[J].Physics Letters A,2007,381:93-104.
Stability of Stochastic Neural Networks with Discrete and Distributed Time-Varying Delays
KANG Wei,LI Lin-guo
(College of iInformation Engineering,F(xiàn)uyang Teachers College,F(xiàn)uyang 236041,China)
The global exponential stability of stochastic neural networks with discrete and distributed time-varying delays is investigated.According to the Lyapunov stability theory and LMIs approaches,delay-dependent criteria are derived to ensure the global robust exponential stability of the addressed system in the mean square for all admissible parameter uncertainties.A numerical example is given to illustrate the effectiveness of the results obtained.
neural network;discrete and distributed time-varying delays;exponential stability;LMIS
book=37,ebook=135
O193
A
10.3969/j.issn.1007-2985.2012.04.008
(責任編輯 陳炳權)
1007-2985(2012)04-0037-04
2012-04-25
安徽省青年人才基金資助項目(2010SQRL196);阜陽師范學院校級資助項目(2011FSKJ14)
康 衛(wèi)(1985-),男,亳州利辛人,阜陽師范學院信息工程學院教師,碩士,主要從事神經網絡及控制理論研
究.