Cohen-Grossberg神经网络稳定性分析  

Stability analysis criterion for Cohen-Grossberg neural networks

作　　者：武怀勤[1] 秦雷杰[1] 石蕊[1] 冯涛[1] 贺丽君[1] 

机构地区：[1]燕山大学理学院,河北秦皇岛066004

出　　处：《辽宁工程技术大学学报（自然科学版）》2011年第3期451-454,共4页Journal of Liaoning Technical University (Natural Science)

基　　金：中国河北省教育基金资助项目(2009157)

摘　　要：为了研究具有逆Lipschitz激励函数的Cohen-Grossberg神经网络的稳定性,应用Brouwer拓扑度性质和线性矩阵不等式技术,探讨了Cohen-Grossberg神经网络的平衡点的存在性及唯一性。通过构造合适的Lyapunov函数和利用Lyapunov对角稳定性矩阵,给出了唯一平衡点全局指数稳定的充分条件。The purpose of this study is to present a novel class of Cohen-Grossberg neural networks with inverse-Lipschitz neuron activation functions. By employing the Brouwer degree properties and linear matrix inequality techniques, the existence and uniqueness of equilibrium point for Cohen-Grossberg neural networks are investigated. With the construction of appropriate Lyapunov functions and the application of Lyapunov diagonally stable matrices; this study provides a sufficient condition which is used to check the global exponential stability of a unique equilibrium point.

关 键 词：神经网络 Cohen-Grossberg模型 逆Lipschitz函数 Brouwer拓扑度 稳定性分析 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]