为了能够方便地分析离散递归神经网络(RNN)的稳定性,以及解决目前比较难的离散非线性系统的控制器的综合等问题,类似于鲁棒控制中的标准模型,提出一种新的神经网络模型——离散时滞标准神经网络模型(DDSNNM),它由离散线性动力学系统和有界静态时滞(或非时滞)非线性算子连接而成.利用不同的Lyapunov泛函和S方法推导出基于线性矩阵不等式(或非线性矩阵不等式)的DDSNNM全局渐近稳定性和全局指数稳定性的充分条件.大多数离散时滞(或非时滞)RNN稳定性分析或包含神经网络的非线性控制系统都可以转化为DDSNNM形式,从而进行稳定性分析或镇定控制.从DDSNNM应用于离散时滞细胞神经网络(CNN)的稳定性分析以及非线性控制系统的综合实例可以看出,DDSNNM不仅使得大多数RNN的稳定性判定简单易行,而且为非线性系统的控制器设计提供新的思路. In order to analyze the stability of discrete recursive neural network (RNN) conveniently and to solve the problem of synthesizing controllers in discrete nonlinear systems which are difficult at present, similar to the standard model in robust control, a new neural Network Model - Discrete Time Lag Standard Neural Network Model (DDSNNM), which is composed of a discrete linear dynamic system and a bounded static delay (or non-delay) nonlinear operator. Using different Lyapunov functional and S Methods The sufficient conditions for global asymptotic stability and global exponential stability of DDSNNM based on linear matrix inequalities (or nonlinear matrix inequalities) are derived.Most discrete time delay (or non-delay) RNN stability analysis or neural network The nonlinear control system can be transformed into DDSNNM form for stability analysis or stabilization control.From the stability analysis of DDSNNM applied to discrete time-delay cellular neural network (CNN) and the comprehensive example of nonlinear control system, it can be seen that DDSNNM Not only makes the stability judgment of most RNNs easy, but also provides a new idea for the controller design of nonlinear systems.