The Position Sensorless Control For PM Brushless DC Motor Based On WNN

Permanent Magnet Brushless DC Motor (BLDCM) has many advantages such as simple structure, reliable, easy maintenance, high operating efficiency and good timing performance, so it is widely used in various fields of national production. Generally, the commutating signal of BLDCM could be got from rotor position sensors. But the existence of rotor position sensors increases the motor's size, costs and reduces its operational reliability. Therefore, it has important value in theoretical and practical to study position sensorless control methods for BLDCM.To get commutating signals from some signals of the motor without position sensor can be regarded as a nonlinear system identification process, and be solved by the identification methods. Artificial Neural Networks (ANN) now has become an excellent nonlinear system identification method for its advantages, such as approach ability, good compute ability and robust. Because Wavelet Neural Networks (WNN) has the merits of fast, accuracy and global convergence, which ANN or Wavelet possesses, it has great potential in nonlinear system identification.This paper concentrates on position sensorless control for BLDCM, proposes a novel control method that used WNN. Firstly, the principle of position sensorless control for BLDCM and some control methods are analyzed in detail. Then the relationship between rotor's position and phase voltages of BLDCM is analyzed, and the direct voltage control method based on WNN for BLDCM without position sensor is subsequently proposed. Adaptive algorithm is used in this paper to choose the number of WNN's hidden units, and the WNN is finally built with a much simpler and tighter structure. Thus enhances the real-time character of networks, and avoids the tendency of unconvergence and mess calculation caused by improper hidden units'number. The parameters of network are initialized by offline training in this paper, while the weights of WNN are updated by online learning. Thus, the approach ability and adaptive capacity are both ensured. At last, the method of direct voltage control based on WNN for BLDCM without position sensor is confirmed by simulation. A control system for BLDCM is designed in this paper based on TMS320F2812, and the method proposed in this paper is experiment on this system. The experimental results show that this method can provide an accurate commutating signal for BLDCM, and is proved to be highly accurate, good self-adaptable, fast dynamic responsibility and robust.