The Speed-Regulating System Of BLDCM With Neural Network

With the development of industry and aviation, the performance and precision of control system are being concerned more and more. The brushless DC motor (BLDCM) control system is a novel speed-variable system. It offers excellent characteristics of operation, control and economy, and shows great developing potentiality. The speed-regulating system of BLDCM is a nonlinear system with high rank and strong coupling. To improve the quality of system, control strategy is very important. In view of above-mentioned reasons, a study on neural network to the speed- regulating system of BLDCM is accomplished in this thesis.Firstly, the basic conception of neural network is introduced systematically. Then, the basic theory of neural network using in control system is analyzed, and the main function of neural network using in control system is summarized.Secondly, the mathematical model of BLDCM is studied. Then, the simulation model of BLDCM is built in MATLAB6.5.Thirdly, for making up the deficiency of traditional PID control, single neural adaptive PID control and BP neural network PID control are used. Then, the research of simulation is accomplished. Thereinto, controller is made up by s-function.Finally, based on the theory of neural self-Tuning control, the non-linear controller is transformed into a linear one by using two neural network to draw up two non-linear function. Neural network is composed of import layer, connotative layer and output layer. Then, the operation principle is analyzed from structure, forward calculation and the adjustment rules of coefficients. Thereinto, in simulation research of error back propagation, the arithmetic of Levnberg-Merqnardt is proved to be the most suitable one after attempting various improved BP arithmetic. The simulation results show that the neural self-tuning controller has better dynamic performance and powerful robustness than traditional PID controller.