Research On The Control Algorithm For Hydraulic APC Of An Experimental Rolling Mill

Automatic Gauge Control (AGC) is particularly critical in the field of metal rolling. AGC System is indispensable in the automatic systems of rolling mill, but for the reason of the complexity, nonlinear and time-varying of AGC it is hard to build an exact mathematical model for it. Therefore, intelligent algorithms are reseashed in this paper, and apply some modern control theories to AGC will improve the control performances and the quality of mill products.This paper takes Hille 100 experimental rolling mill as the research object. The components and working principles of AGC system were introduced. And the transfer functions for the main components of APC system were analyzed. Finally the mathematical model of the APC system was established.The basic principles and methods of parameters tuning for PID control were intrduced in this paper. A conventional PID controller was designed based on Ziegler-Nichols method according to the mathematical model of the APC system. It is difficult to achieve good control performances with PID control for the practical system is nonlinear and time-varying, so it's necessary to use intelligent algorithms to improve the ability of parameter tuning for PID. Therefor, neural network and fuzzy control were studied in this paper and single neural PID controller, BP neural network PID controller and fuzzy PID controller were designed. Simulation results show that the three controllers have good control performances, with better robustness and adaptability compared to conventional PID controller. Since the establishment of fuzzy control rules was based on expert knowledge and operator experience, it is hard to achieve the optimal control performance for fuzzy PID controller, therefore, genetic algorithm was introduced to optimize the fuzzy control rules in this paper and it improved the control performance further.In order to verify the control algotithms, this paper designed a motor control system as an application and genetic algorithm was adopted to identify the model of the system. Subsquently, a PID controller and fuzzy PID controller were designed based the model identified and both applicated successfully. The results show that the model identified is suitable, and fuzzy PID has better control performance compared to conventional PID. This paper provides a reliable reference for the design and optimal of controller in pratical industrial.