Permanent Magnet Synchronous Motor System Based On Neural Network Self-turning Speed Control

Permanent magnet synchronous motor(PMSM) has attracted many researchers’ attention and been widely applied to modern AC variable speed system, because the the advantages of small volume, light weight, stable speed, fast dynamic response, simple control torque and strong overload capacity and some others. However, the PMSM is a nonlinear, multivariable and strong coupling control system, the traditional PID speed controller are susceptible to some uncertain factors such as motor parameter variations and load disturbances, So it has been difficult to satisfy the high control performance of PMSM system under the condition of high precision of speed control and large load fluctuation.Neural network is often used to solve the complex problem of nonlinear control system because it has self-learning and adaptive ability. With the mature development of neural network control theory, neural network has been more and more widely used in the field of motor control. BP neural network is a earlier control method in the field of neural network, but It is difficult to meet the requirements of real-time control system because the slow training rate and poor generalization ability. On the contrary, by adopting appropriate training algorithm, the RBF neural network can achieve fast learning rate, good generalization ability and high real-time requirements. Therefore, this thesis focus on studying permanent magnet synchronous motor speed control method based on RBF neural network self-turning PID to achive high performance control of PMSM vector system.This thesis combine RAN learning algorithm and the gradient descent methods to train the RBF neural network to get a network structure which have fast learning speed and good dynamic performance. And then through the self-learning and self-adaptive ability of the trained RBF neural network to adjust the parameters of PID contoller,achieving the high performance speed control of PMSM.In order to verify the feasibility of the scheme, this thesis established the RBF neural network self-tuning of the permanent magnet synchronous motor vector control system simulation model at load disturbance and speed change case on MATLAB/SIMULINK simulation platform, and analysis of the control performance with compare to conventional PID speed control. The simulation results show that the RBF neural nerwork self-turning PID controller is been affected small under the variation of motor parameters and load disturbances, and has good adaptability and robustness, can be quickly and stablely for speed tracking control.Finally, design and build the experimental platform of the vector control system of permanent magnet synchronous motor with TI company’s TMS3202812 as control chip, and do some relevant experiments. Experimental results furtherly validate the feasibility and superiority of the proposed RBF neural network self-turning PID speed control of permanent magnet synchronous motor vector control system.