Universal learning network predictive control for nonlinear dynamic systems with time-delay

In this paper, universal learning network (ULN), is used to identify the typical nonlinear and long time delay system. Integrating ULN with a neural-PID controller, this paper achieves an exact predictive control for pH neutralization process. Compared with the traditional Smith predictive control method, it is proved that the general architecture and learning algorithm give ULN more representing abilities to model and control the nonlinear complex systems with long time delay. It is a new effective method for identification with unknown object model.