Study On Shape Control Of HC Mill Based On Fuzzy Neural Network

The shape control technology is the key technology in the modern strip production. The development of the shape theory has produced a great effect on the strip rolling technique. But with the increasing demands of the quality of rolling strip production in the market, the weakness has evidently appeared in the conventional theory. With the improvement of computer, the artificial intelligence has been widely applied in different fields. The combination of the artificial intelligence technique and the rolling technique has been made in the rolling theory and has achieved great results. Taking the 900HC cold strip rolling mill as an object of study, a research for the application of the theory of fuzzy neural network control has been made to the shape control in the cold rolling mill. The relationship of the means of adjusting and the error of shape is firstly established in the study , and then the characteristic of tilting rollers and bending rollers is obtained on the basis of conclusion. Through the fuzzy inference and analysis of the object of being control and the conclusion of the experience of the operators, the control rules and study sample of the controller have been gained. A fuzzy neural network controller based on BP algorithm has been designed, and through the off-line training, adjusting the membership functions and memorizing the fuzzy rules, then through the on-line automatic adjusting, further improving the effect of control and realizing the self-adaptation of the process of control. Finally, the simulation research to rolling process is made by using this kind of fuzzy neural network controller and the result of simulation proves that the controller can better correct the error of the shape of plate and also fully proves the feasibility of the method, which improves the quality and accuracy of the shape control.