The Research Of Fuzzy Neural Network Control System Of BLDCM

Brushless DC motor has the features of time varying, linearity and strong coupling. Although traditional control methods are of easy arithmetic, fast performance and precise control, it is difficult to meet the needs of static state and dynamic performance when the model of control subjects is uncertain. Intelligent control does not depend on exact math models and can restrain the impact of time-varying and parameter disturbance, so the intelligent control and the traditional control can complement each other.This thesis designs a fuzzy neural network (FNN) controller in order to resolve the problems above. FNN is a new type of neural network, which introduces fuzzy arithmetic or fuzzy weights. FNN can combine the fuzzy logic arithmetic and neural network together, which either has the merit of easy fuzzy control expression and the good self-learning capacity of neural network or adjusts the rules and the parameters of controller according to the variation of parameters of the control subject and the environment. So neural network and fuzzy control can complement each other, raise the overall the capacity of learn and expression, expand their respective application domains, and form a more efficient knowledge processor.However, during the course of design, there are some problems on the structure and learning arithmetic of FNN. The thesis offers some proving methods and theory deduction in detail as well as the simulation results to demonstrate its control feature. Regarding real-time and the realization of micro-processor, the paper puts forward Kalman filter arithmetic and proves it. After the simulation of MATLAB, it is evident that the FNN adopting Kalman filter uses less learning time and restrains the noisy disturbance.After we analyzing the model of BLDCM, the FNN control which adopts Kalman filter arithmetic is applied in the speed controller. The simulation results demonstrate that the FNN control system of BLDCM has good static state and dynamic performance, and reduces torque tremble to some degree. In short, the control results can satisfy us.