The Study To Improve The Prediction Of The Rolling Load In The Finishing Train Of 1700 Hot Strip Mill Of The Wuhan Iron&Steel Co.

One of the main functions of the control systems is the setup of the mills, and the prediction of the rolling pressure is the most important part, which is the key of the control system. The result of the prediction will affect many factors such as the setup of the rolling gap and the quality of the finished products etc. Facing with the increasing need of the consumers, it is playing a more and more important role in the industrial process under the heating market competition. The process of rolling is very complicated and the number of linear and non-linear parameters that determines the final properties can be quite large. Therefore, it is extremely difficult to develop a physical model for prediction. Because the result of the model using in the Wuhan Iron&Steel Co. is not very ideal, so it is need to do a further improvement.In this paper I selected two kinds of carbon steels (WSPCC and Q235B) as sample from Wuhan Iron&Steel Co. to do a study of the prediction of rolling pressure using the following four techniques. 1, Using the traditional model to predict directly; 2, Using the traditional non-linear and polynomial regression techniques to obtain a new mathematical model of rolling pressure to predict; 3 > Using the Back-Propagation neural networks to predict directly; 4> Combining the Back-Propagation neural networks and the mathematical models to predict rolling pressure. Off-line prediction indicates that the predicted result of combining the Back-Propagation neural networks and the new mathematical models obtained is the most ideal. Especially the improvement on the prediction of the finishing stand is obvious. For the two kinds of carbon steels I selected, the rate of the samples that the relative error above 10% is dropped from 62.3% and 45.6% to 8.8% and 16.7%.