Extended Dissipativity Analysis of Delayed Memristive Neural Networks Based on A Parameter-Dependent Lyapunov Functional

Abstract

This paper is concerned with extended dissipativity analysis of memristive neural networks with time-varying delays. Using the characteristic function technique, a tractable model of a memristive neural network is obtained. This model is similar to a neural network with polytopic uncertain synaptic weights, enabling us to construct a parameter-dependent Lyapunov functional. By combining this functional and some integral inequalities, a novel extended dissipativity criterion is obtained in terms of linear-matrix-inequalities, where different Lyapunov matrices are used for each form of the memristive neural network. Through a numerical example, this criterion is shown to be less conservative than the one based on a common Lyapunov functional.