| Fuzzy control and neural network are important branch of intelligent control., and genetic arithmetic is a excellent algorithms of optimization in field of artificial intelligence. It's a active research in current domain of intelligent control that how to integrate them with reason.This paper discusses some aspects of structure changed neuro-fuzzy controller which is applied in temperature control for a rubber drying furnace. For simple fuzzy controller, picking up appropriate fuzzy rules from experts' knowledge and experience inevitably includes some subjectivity so that quality of control is limited, at a certain extent Structure changed neuro-fuzzy control system integrate neural network and fuzzy control, pick up subset of fuzzy rules from all possible fuzzy rules, achieve quick, full-scale optimization of structure and parameters by genetic arithmetic, and improve shortcomings of BP arithmetic, so trained controller has good quality of control. And moreover, structure changed neuro-fuzzy controller has sound ability of resistance to random disturbance. The results of computer simulation demonstrates that this method can achieve satisfactory dynamical characteristic, static characteristic and good robusticity. It has a advantage over traditional PID and fuzzy control, and is applicable to objects that have not (or is difficult to) achieve exact model.Rubber drying furnace has large inertia and nonlinear system, and it's difficult to achieve its exact mathematic model. Because it is controlled by hand, range of change of temperature is quite wide so that it affects quality and quantity of rubber. This paper designs the ad hoc software and hardware of computer control system, The design of hardware is reasonable, economical, and efficient to disturbances. And software has rational structure of modules and satisfactory control effects. Its Operation is simple because of friendly man-machine interface. And it can be regarded as reference for similar system... |
PDF Full Text Request