Dynamic Behavior Analysis Of Several Neural Networks With Time Delay

As a mathematical model for information processing,neural network is one of the most active branches of computational intelligence and machine learning,and has flourished in the fields of secure communication,associative memory,combinatorial optimization and so on.Notice that most of these achievements are based on the fact that neural networks have a desirable dynamical behavior.However,the existence of time delay may make the neural network towards the undesirable dynamic behavior.Therefore,it is of practical significance to study the dynamic behavior of neural networks with time delay.This paper focuses on the dynamic behavior analysis of several kinds of neural networks with time delay.By means of average impulsive interval,comparison principle,average impulsive delay,differential inclusion principle and linear matrix inequality technique,combined with specific neural network,the dynamic evolution process is analyzed,and some theoretical criteria of dynamic behavior are obtained.The main work of this paper is as follows:The time-delay effect on the stability of a class of impulsive neural networks is studied.Aiming at the interference of the delay term in continuous dynamics to the impulsive system,the sufficient conditions for the stability of the delayed impulsive neural network are obtained by using the concept of average impulsive interval.A strict comparison principle is proposed to prove that the impulsive system can maintain the original stability under certain conditions for any large but bounded delay.In particular,as an extension,the stability of delayed impulsive neural networks containing both stabilizing and destabilizing impulses is also discussed.The synchronization of uncertain chaotic neural networks with constant delay is explored.With the help of the average-delay impulsive control method and the newly proposed impulsive delayed inequality,the synchronization criteria of chaotic neural networks based on norm bounded uncertain parameters are derived.The problem of periodic intermittent event triggering stabilization for a class of time-delay memristive neural networks with uncertain parameters is investigated.By introducing event-triggered strategy into periodically intermittent controller,some sufficient criteria to ensure the global asymptotic stability of the system are established,and linear matrix inequalities are used to incorporate them into controller design to determine the controller gain and the triggered parameters.In addition,it is verified that the proposed event-triggered strategy avoids the Zeno behavior.