变时滞神经网络的时滞相关全局渐近稳定新判据  

A new delay-dependent global asymptotic stability for neural networks with time-varying delays

作　　者：刘国彩[1] 刘玉常[1] 鞠培军[1] 

机构地区：[1]泰山学院数学与系统科学系,山东泰安271021

出　　处：《山东大学学报（工学版）》2010年第4期53-56,61,共5页Journal of Shandong University（Engineering Science）

基　　金：山东省教育厅科研发展计划资助项目(J07WJ23)

摘　　要：研究了变时滞神经网络的时滞相关渐近稳定性问题。利用Lyapunov-Krasovskii泛函和积分等式方法,提出了此类神经网络有唯一稳定点和全局渐近稳定的充分条件。与已有文献相比,不再要求激活函数单调,并且时滞导数可以不小于1,具有较弱的保守性。算例验证了所给方法的正确性。The asymptotic stability of neural networks with time-varying delays was studied.By the use of the Lyapunov-Krasovskii functional and integral-equality approach,sufficient conditions were developed to ensure that the delayed neural network has a unique equilibrium point and global asymptotic stability.Monotone was not required by the activation function,and the delay-derivative could be 1 or more than 1.Reduced conservativeness of the proposed method was verified with an example.

关 键 词：神经网络 积分等式 全局渐近稳定 线性矩阵不等式 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]