Strip thickness control of reversing mill using self-tuning PID neurocontroller

A self-tuning PID control approach is presented for improvement of the head and tail strip thickness accuracy in a reversing cold mill for offering a cost saving. A neural network is used on-line to tune the parameters of a conventional PID controller in AGC to improve the response of strip thickness during a transient rolling process, which results in a reduction of off-gauge strip length. The effectiveness of the presented approach has been demonstrated through a simulation example. The results of simulation show that a neural network can reduce the strip thickness error quickly during mill starting process while the PI controller parameters are bei ng tuned on-line, so that a saving of off-gauge strip length about 73% is achieved.