李蕾 何秀麗
摘要
通過向量Lyapunov函數,給隨機CGNNs以均方估計,研究基于馬氏切換的脈沖時滯隨機CohenGrossberg神經網絡模型的均方指數穩定性,并利用數值例子對結論加以證明.關鍵詞CohenGrossberg網絡模型;均方指數穩定性;馬氏切換
中圖分類號O175
文獻標志碼A
0引言
在過去的幾十年里,神經網絡在各個領域有著廣泛的研究和應用,吸引了國內外許多學者的關注[15].CohenGrossberg神經網絡模型,由Cohen和Grossberg在1983年首次提出[1],包括著名的細胞神經網絡模型、Hopfield網絡模型(HNNs),以及作為其特殊情況的LotkaVolterra競爭生態模型(LVCMs).因為其在各領域的廣泛應用,如聯想記憶、模式分類、并行計算、機器人、計算機視覺和最優化等,近幾年被研究人員廣泛研究和引用.
時間延遲、脈沖擾動是導致神經網絡不穩定的因素.在現實生活中,時滯對于神經網絡的研究來說是不可避免的,是CGNNs頻繁振蕩和不穩定的來源,所以研究時滯CGNNs的穩定性具有重要的意義.Xu等[2]研究討論了時滯隨機CohenGrossberg網絡模型的均方穩定性.另一方面,脈沖也是必不可免的,脈沖能使穩定的系統不穩定或者使不穩定的系統穩定.它應用在各個領域,如生物學、種群系統等.因此考慮脈沖作用下時滯隨機神經網絡系統的均方指數穩定性是很有必要的.越來越多的研究開始集中在脈沖神經網絡和脈沖時滯隨機神經網絡的穩定性分析,并取得了一些重要成果[34].
最近幾年研究的脈沖神經網絡模型大多基于標量算子穩定性分析[513],基于向量算子脈沖神經網絡穩定性分析的研究很少,例如周偉松等[14].所以基于向量算子研究脈沖CGNNs的均方指數穩定性已成為一個具有重要的理論和實踐意義的課題.本文通過在特定時刻添加脈沖干擾,將L算子以及伊藤公式結合起來應用到CGNNs,來研究帶有馬氏切換的隨機脈沖CohenGrossberg神經網絡模型的均方指數穩定性.
1預備知識
4討論
穩定性不僅是神經網絡應用的基礎,同樣也是神經網絡最基本和重要的問題.近年來,有不少學者對隨機神經系統的穩定性進行了大量的研究和應用.在此基礎上,得到了隨機脈沖時滯系統保持穩定性的條件.研究帶有馬氏切換隨機脈沖時滯CGNNs的均方指數穩定性突破了傳統只研究沒有時滯的隨機CGNNs的局限性,通過使用Halanay不等式以及伊藤公式得到了系統均方指數穩定性的充分條件.所討論的隨機脈沖時滯CGNNs不僅在理論上有著廣泛的研究,在實際上也有著很大的發展前景.
參考文獻
References
[1]Cohen M A,Grossberg S.Absolute stability and global pattern formation and parallel memory storage by competitive neural networks[J].IEEE Transactions on Systems,Man,and Cybernetics,1983,13(5):815826
[2]Xu D Y,Zhou W S,Long S J.Global exponential stability of impulsive integrodifferential equation[J].Nonlinear Analysis(Theory,Methods & Applications),2006,64(12):28052816
[3]Yao F Q,Cao J D,Cheng P,et al.Generalized average dwell time approach to stability and inputtostate stability of hybrid impulsive stochastic differential systems[J].Nonlinear Analysis(Hybrid Systems),2016,22:147160
[4]Xu D Y,Yang Z C.Impulsive delay differential inequality and stability of neural networks[J].Journal of Mathematical Analysis and Applications,2005,305(1):107120
[5]Luo T Q,Long S J.A new inequality of Loperation to stochastic nonautonomous impulsive neural networks with delays[J].Advances in Difference Equations,2016,DOI:101186/s136620150697y
[6]Song Y F,Shen Y,Yin Q.New discrete Halanaytype inequalities and applications[J].Applied Mathematics Letters,2013,26(2):258263
[7]Tang C Q,Wu Y H.Global exponential stability of nonresident computer virus models[J].Nonlinear Analysis:Real World Applications,2017,34:149158
[8]Zhang Y.Global exponential stability of delay difference equations with delayed impulses[J].Mathematics and Computers in Simulation,2016,132:183194
[9]Sun F L,Gao L X,Zhu W,et al.Generalized exponential inputtostate stability of nonlinear systems with time delay[J].Communications in Nonlinear Science & Numerical Simulation,2016,44:352359
[10]王慧.脈沖時滯神經網絡的全局穩定性研究[D].重慶:重慶大學計算機學院,2007
WANG Hui.Study on global stability of impulsive delayed neural networks[D].Chongqing:College of Computer Science,Chongqing University,2007
[11]彭國強.隨機神經網絡的穩定性[D].長沙:湖南大學數學與計量經濟學院,2009
PENG Guoqiang.Stability of stochastic neural networks[D].Changsha:College of Mathematics and Econometrics,Hunan University,2009
[12]Shen Y,Wang J.Almost sure exponential stability of recurrent neural networks with Markovian switching[J].IEEE Transactions on Neural Networks,2009,20(5):840855
[13]Li Z H,Liu L,Zhu Q X.Meansquare exponential inputtostate stability of delayed CohenGrossberg neural networks with Markovian switching based on vector Lyapunov functions[J].Neural Networks,2016,84:3946
[14]Zhou W S,Teng L Y,Xu D Y.Meansquare exponentially inputtostate stability of stochastic CohenGrossberg neural networks with timevarying delays[J].Neurocomputing,2015,153:5461
[15]Wang X H,Guo Q Y,Xu D Y.Exponential pstability of impulsive stochastic CohenGrossberg neural networks with mixed delays[J].Mathematics & Computers in Simulation,2009,79(5):16981710
AbstractFocused on CohenGrossberg neural networks,this paper investigates the meansquare exponential stability by means of the vector Lyapunov function.This method ensures that the impulsive stochastic CohenGrossberg neural network is exponentially stable.Finally,an example is used to illustrate the conclusions.
Key wordsCohenGrossberg networks; meansquare exponential stability; Markovian switching