Study On Neural Networks By Intermittent Control

The neural network as a kind of emerging information Processing science can abstracts and simulates some basic characteristic of the human brain. It is an information processing method that is auto-adapted and non-procedural, taking person's cerebrum working pattern as a foundation. The characteristic of this kind work mechanism is to display its own processing functional by the massive neurons in the network. It is by simulating the human brain structure and the single neuron function to achieve the goal that simulates the human brain process information.At present, in the national economy and modernization of national defense science and technology, the neural network has the broad application and the development prospect in information, automated, project, economical and so on. During the past few years, the problem of stability of neural networks has been one of most active areas of research and has attracted much attention.In this dissertation, the intermittent control method will be introduced and then applied to chaotic neural networks. As a special form of switching control, intermittent control is a direct and efficient engineering approach for any type of process control. It has been used for a variety of purposes in engineering fields such as manufacturing, transportation, air-quality control, and communication. In the third chapter, we will formulate the exponential stabilization problem for a class of delayed chaotic systems by means of periodically intermittent control. An exponential stability criterion for the controlled neural networks, together with its simplified version, is established by using Lyapunov function and Halanay inequality. The feasible region of control parameters is estimated in a rigorous way. The theoretical results and the numerical simulations show that the continuous-time delayed chaotic neural networks can be stabilized by intermittent feedback control with nonzero duration. In the fourth chapter, we describe a method for the weak synchronization of chaotic systems with parameter mismatch by using periodically intermittent control. By this method, a proper value of control parameter can be obtained and the synchronization error converges exponentially to a small region.