双向联想记忆神经网络的指数输入-状态稳定性  

Exponential Input-to-State Stability of Bidirectional Associative Memory Neural Networks

作　　者：李建军[1] 杨志春[1] 

机构地区：[1]重庆师范大学数学科学学院,重庆401331

出　　处：《重庆师范大学学报（自然科学版）》2016年第4期79-84,共6页Journal of Chongqing Normal University:Natural Science

基　　金：国家自然科学基金(No.11471061);重庆市自然科学基金(No.CQCSTC2014JCYJA40004);重庆市高校创新团队计划(No.KJTD201308);重庆市研究生科研创新项目(No.CYS16149)

摘　　要：研究关于具有多个时滞效应和时变外部输入双向联想记忆神经网络模型的指数输入-状态稳定性分析。首先,建立了双向联想记忆神经网络模型,该模型具有多个时滞效应并且外部输入是时变的。而且模型中非线性神经元激励函数不要求是有界的,也不要求是光滑的。然后给出双向联想记忆神经网络指数输入-状态稳定性的一个定义,利用Lyapunov泛函和线性矩阵不等式-mX^TQX+2lX^TπY≤l2Y^Tπ~T(mQ)^(-1)πY和X^TY+Y^TX≤εX^TΛX+ε\Y^TΛ^(-1)Y的方法,获得含有多时滞效应和时变外部输入的双向联想记忆神经网络模型指数输入-状态稳定性的一个充分条件。We study exponential input-to-state stability of bidirectional associative memory neural networks with multiple time delays effects and external input with time varying. Firstly, a model of BAM neural network is established, in which there are multiple time delay effects and the external inputs with time varying. We don＇t require that active functions are bounded and smooth in the BAM networks. Then, the definition of exponential input to-state stability for BAM neural networks is given. We obtain a sufficient condition ensuring exponential input-to-state stability of BAM neural networks with multiple time delays and external input with time varying by using Lyapunov function method and the linear matrix inequality-mX^TQX＋2lX^TπY≤l^2Y^Tπ^TmQ）-1πY和X^TY＋Y^TX≤εX^TΛX＋ε／Y^TΛ-1Y.

关 键 词：时滞 双向联想记忆神经网络 指数输入-状态稳定 LYAPUNOV泛函 

分 类 号：O175[理学—数学]