单机架炉卷轧机监控AGC控制算法的研究

Research of monitor AGC arithmetic in single-stand Steckel mill

为解决单机架炉卷轧机监控AGC的大时滞和控制精度问题,针对常规PID不能自我进化的缺点,本文采用BP神经网络对PID控制器进行参数自整定,并将该智能控制器与Smith预估器结合对AGC系统进行监控,既解决了自我进化问题,也克服了时滞,通过仿真分析和现场验证,其收敛速度和鲁棒性明显改善。

Considered the general PID can＇t improve itself, a BP neural network PID is proposed to improve the control precision and large time deday of monitor ACC of a Steekel rolling mill. And then the intelligent PID and Smith prediction control machine are combined to minitor the A（;C system so that the self-learning capability and time delay of the system evidently improved. It was proved that the robustness and convergence rate of the system is obviously improved by simulation and implement experiment.