基于神经网络的永磁同步电动机转矩脉动补偿

永磁同步电动机的转矩波动补偿多采用转矩估计方法，但在不同电机运行点和电机参数不断变化的情况下，准确估计瞬时转矩非常困难．因此提出了采用径向基函数神经网络作为转矩波动补偿器的永磁同步电动机伺服控制方法．利用神经网络在线逼近非线性因素和外部干扰，对转矩波动进行补偿，然后根据反步控制方法得到神经网络权值调整规则和控制器输出．采用数字信号处理（DSP）进行了实验研究，结合神经网络和Backstepping的优点实现永磁同步电动机的转矩波动补偿．仿真和实验结果证明，采用上述方法可以实现不同转矩脉动的补偿。

Torque Ripple Minimization in Permanent Magnet Synchronous Motors Based on Neural Networks

The method of torque estimation is generally adopted for compensating the torque ripple of permanent magnet synchronous motors(PMSM).But at different points and with changing motor parameters,precise estima- tion of the instantaneous torque is difficult.Therefore,a method is proposed,which uses radical basis function neural networks(RBF NN)as the torque ripple compensation to control the PMSM.First,the torque ripple is compensated by on-line approximating the nonlinear factors and external disturbances with neural networks (NN),and then by Backstepping control method the rules of modulating the weights of NN and the output of the controller are obtained.Simulation and digital signal processing(DSP)experiment results show that the proposed method combines the advantages of both NN and Backstepping and can realize different kinds of torque compen- sation.