High Performance Sensorless Permanent Magnet Synchronous Motor Direct Torque Control

With the rapid development of power electronics, microelectronics, new motor control theory and the rare earth permanent magnet materials, permanent magnet synchronous motor get rapid popularization and application. Because of its low loss, high efficiency, small size and other advantages, permanent magnet synchronous motor has been widely used in various fields of national life and production. The development and maturing of sensorless technology solve a series of problems caused by using conventional mechanical sensors in permanent magnet synchronous motor. It not only simplifies the motor structure and lowers the cost, but also improves the robustness of the motor system.This paper focuses on the study of sensorless control of permanent magnet synchronous motor. In the step of sensorless speed identification, a new method is put forward on the basis of the analysis of BP neural network, wavelet transform and wavelet neural network method, which is using particle swarm optimization algorithm instead of the conventional wavelet neural network gradient descent method and wavelet neural network is constructed based on particle swarm optimization. Experiments show that the wavelet neural network based on particle swarm optimization algorithm solves the problem that the conventional wavelet neural network is likely to fall into local minima, and has a fast convergence in network training with a higher precision.Two of the most popular control methods are compared in this paper which are vector control and direct torque control. After analyzing the existing shortcomings of traditional direct torque control, improvements are made, which are adding the number of magnetic linkage spatial location from six to twelve and using fuzzy controller instead of torque hysteresis comparator and magnetic linkage hysteresis comparator. By refining the switch selection table, more appropriate voltage space vector can be selected to adjust torque and stator magnetic linkage for the motor. The simulation experiments on the MATLAB platform proved that 12-vector dividing and using fuzzy controller can effectively restrain the pulsating movement of magnetic linkage and torque, and improve the controlling performance of the motor.