Research On Sliding Mode Control For Linear Permanent Magnetism Synchronous Motors Based On Neural Network Optimization

Sliding mode control has the characteristic of celerity, robust, and easy to implementation, but the chattering exists in the sliding mode. Neural network control strategy has the abilities of self-learning and adapt to the system uncertainties. For linear permanent magnetism synchronous servo motors (LPMSM), due to the cause of chattering, using the neural network-based compensation methods are proposed to optimize the sliding mode control and weaken the system chattering, so that the system could rapidly and accurately track to the command signal. First, using the MATLAB/SIMULINK, a general simulation model of LPMSM controlled by SPWM vector current sluggish tracking is established with block diagram through the analysis and research of LPMSM and its vector control principles. Then the thesis introduce three methods of optimizing the sliding mode control by the neural network: a two-degree-freedom controller with neural network-sliding mode control, sliding mode controller's celerity could keep the track capability of the system, and designing the feedback controller using neural network to restrain the influence of parameter variations and disturbance, and weaken the chattering. A sliding mode controller based on neural network observer, because the linear observer has weak adaptability to parameter variations, we designing a neural network observer parallels to linearobserver, then using the summation to come true compensate. A position tracking sliding mode controller based on RBF neural network compensation, using RBF neural network learning the uncertainties and disturbance, so that compensates real time and weaken the effect of chattering to system.At last, the simulation research result which based on the general simulation model of LPMSM, show the effectiveness of the control strategies that designed in this paper.