Improved Smith prediction monitoring AGC system based on feedback-assisted iterative learning control

ZHANG Hao-yu(张浩宇), SUN Jie(孙杰), ZHANG Dian-hua(张殿华), CHEN Shu-zong(陈树宗), ZHANG Xin(张欣)

(State Key Laboratory of Rolling and Automation, Northeastern University, Shenyang 110819, China)

Abstract:The performance of Smith prediction monitoring automatic gauge control (AGC) system is influenced by model mismatching greatly in strip rolling process. Aiming at this problem, a feedback-assisted iterative learning control strategy, which learned unknown modeling error by using previous control information repeatedly, was introduced into Smith prediction monitoring AGC system. Firstly, conventional Smith predictor and improved Smith predictor with PI-P controller were analyzed. Secondly, on the basis of establishing of feedback-assisted iterative learning control strategy for improved Smith predictor, process control signal update law and control error were deduced, then convergence condition of this strategy was put forward and proved. Finally, after modeling the automatic position control system, the PI-P Smith prediction monitoring AGC system with feedback-assisted iterative learning control was researched through simulation. Simulation results indicate that this system remains stable during model mismatching. The robustness and response of monitoring AGC is improved by development of feedback-assisted iterative learning control strategy for PI-P Smith predictor.

Key words:automatic gauge control; Smith predictor; monitoring automatic gauge control (AGC); feedback-assisted iterative learning control; automatic position control