A self-learning method for improving the hot-rolled plate shape

摘要

In the multi-standard rolling process of multi-steel production lines, frequent changes in specifications lead to slow self-learning and inaccurate learning, which makes the index of thickness deviation on plate out of target. To solve this problem, a quick self-learning method was designed and proposed. By calculating the crown deviation value of the current rolled steel strip, the bending roll force variation value and the plate shape feedback compensation value. The control parameters in the model are compensated accordingly to compensate for control deviations caused by changes in rolling conditions and the multi-standard steel. Experiments show that the rapid self-learning method can effectively improve the shape index.