基于小波神经网络的SVPWM算法研究

Research on Space Vector PWM Based on Wavelet Neural Network

分析了SVPWM的基本原理,提出一种小波神经网络空间矢量脉宽调制(WNN-SVPWM)算法,并与BP神经网络空间矢量脉宽调制(BP-SVPWM)做了对比研究。仿真结果表明,网络隐层神经元个数相同时,小波网络比BP神经网络具有更快的收敛速度和更高的误差精度,使用WNN-SVPWM算法控制的永磁同步电动机具有更小的定子电流谐波畸变率和脉动转矩。

The basic principle of space vector pulse width modulation （SVPWM） was analyzed, and a algorithm of SVPWM based on wavelet neural network （WNN-SVPWM） was proposed. Comparison between BP neural network space vector pulse width modulation （BP-SVPWM） and WNN-SVPWM was carried out. Simulation results show that wavelet neural network is superior to BP neural network on convergence rate and error precision, and the permanent magnet synchronous motor driven by WNN-SVPWM has less total harmonic distortion of stator current and pulsating torque than that driven by BP-SVPWM.