Research On Intelligent Methods For Shape Control And Shape Measurement

With the development of automobile and home appliance industry, more and more high-quality strip is in need of. There are two yardsticks, gauge and shape, to weigh the quality of strip. While gauge control technology has been made a rapid progress, there is still a large room for improvement in shape control. Because of severe operating environment and complicated mechanism in rolling process, it's hard for classic methods to get desirable performance. So the research on intelligent methods which are based on neural network and fuzzy logic have become a focus today. In view of shape control problem, and analyzing them in depth, some work about shape control and shape detection or measurement have been done, which is as follows. (1) A predictive model has been developed which is based on regularization method. Through the simulation on measured data, it shows the model has desirable generalization ability and predictive accuracy. (2) In view of time variability, nonlinear character, severe disturbance in bending system, for example in UC rolling mill bending system, an AFNNC(Adaptive Fuzzy Neural Network Controller) has been developed. Compared with routine PID controller, the simulation proves that this controller is desirable to solve the problem of parameter alteration in hydraulic bender, it also improves the anti-interference ability greatly of the control system. Especially it reduces the degree of dependence of analytical models. Finally, the trend of shape control and shape measurement technology is described in the thesis.