Closed-Loop System Identification Based On Relay Feedback And The Application In Model Predictive Control

In recent years, closed-loop system identification and predictive control have always been the research focus in automatic control field. However, in the field of industrial process control, about85%loops still uses PID control strategy. Predictive control in industrial process is not widely applied. There are two main reasons:One hand, the dynamic model which is achieved by system identification is not precise enough, the process future deviation is not predicted accurately through utilizing the forecast models. It leads to actual effect of predictive control strategy is worse than PID control strategy; On the other hand, whether the Dynamic Matrix Control or Generalized Predictive Control, there are many parameters to be adjusted, complex parameter to be selected. The control algorithm contains a high-dimensional matrix inversion. The amount of calculation is large. It’s difficult to implement predictive control in engineering.This paper base on the above two points. In single variable model identification, the relay feedback closed loop which is suitable for the industrial process is utilized. Model parameter identification formula of first order plus dead time model is improved after a lot of trail analysis and reasoning. Under saturation relay feedback, model parameters of the process can be identified accurately through improved identification formula. Then single relay feedback is promoted up to variable relay feedback. Equivalent transfer function of all the loops can be achieved through decentralized decoupling. This reduces the coupling between the loops. Because of simple control algorithm and small online computation, the predictive function control strategy is suitable for industrial process. Utilizing the relay feedback identification results, predictive function controller is designed in Siemens distributed control system APACS+. The simulation results show that both single variable system and multi-variable system, predictive functional control strategy can achieve better performance compared to PID control strategy. It has a certain value in engineering.