分布时滞细胞神经网络的代数抗饱和补偿设计  

Algebraic anti-windup compensation design for a class of distributed time-delayed cellular neural networks

作　　者：何汉林[1] 徐文巍 严路[1] 

机构地区：[1]海军工程大学理学院,湖北武汉430033

出　　处：《系统工程与电子技术》2017年第6期1341-1346,共6页Systems Engineering and Electronics

基　　金：国家自然科学基金(61374003);海军工程大学基础研究基金(20161475)资助课题

摘　　要：针对一类分布时滞神经网络系统提出一种代数抗饱和补偿设计。在不考虑系统输入饱和的前提下,通过系统参数、M矩阵理论、迪尼导数设计出了一种代数判据。在此基础上,当发生输入饱和时,运用舒尔补引理将二次型矩阵不等式(quadratic matrix inequality,QMI)转化成线性矩阵不等式(linear matrix inequality,LMI)。同时给出优化吸引域估计的方法。仿真结果验证了方案的可行性和有效性。An algebraic anti-windup compensation design for a class of distributed time-delayed cellular neural networks is considered. Based on system parameters, M-matrix theory, Dini derivative, an algebraic condition of control for distributed time-delayed cellular neural networks is presented without considering its input saturation. When input saturation is considered, the quadratic matrix inequality （QMI） control criterion which can be changed into linear matrix inequality （LMI） criterion can be gotten based on Schur complement principle. And then the attraction domain and its optimization are presented. The simulation results verify the effectiveness and feasibility of the proposed method.

关 键 词：抗饱和 分布时滞 细胞神经网络 吸引域 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]