Neural Network Control Of Delay Systems

Time-delay is a very common characteristic in modern industry production processes. It makes the controlled parameters unable to reflect the disruption to the system. This kind of process will have a larger overshoot and longer adjusting time which make the control problem more complicated. So time-delay is recognized as one of the most difficult dynamics linked to control. The research of control problem for a time-delay system has great theoretical and practical significance.As a new control method in the 20th century, Neural Network provides new idea to the traditional control problem. Neural Network has many merits. Its special nonlinear mapping ability, self-learning and adaptive ability, information integration ability and good fault tolerance make it very suitable for the modeling and control of complex system. Especially, when the system has uncertain factors, Neural Network shows its superiority. So Neural Network has great significance to the control of a system with time-delay.This thesis provides the identification and several common control methods for a time-delay system. First, it uses BP Neural Network to identify linear and nonlinear time-delay system and analyzes the characteristics of Neural Network Identification. It also shows the identification effect of Neural Network. Then, it analyzes the classical and Neural Network PID control methods which are used in the control of time-delay system. The control effects of these two methods are provided. After the comparison, we find that the control performance of Neural Network PID method is better than classical PID control method. In the last two chapters, the IMC and Smith control methods are analyzed. The control effects of the classical method and the Neural Network method are also compared. Simulation results show that the Neural Network method has better control performance than the classical control method.