基于负载测试的神经网络SPMSM参数在线辨识

Online parameter estimation of neural network PMSM based on load testing

针对表面式永磁同步电机(SPMSM),提出了一种基于负载测试的自适应线性元件神经网络(ANN)参数在线辨识方法。该方法通过在d轴电流上添加一个△id负载信号和对转速做一小范围的变化,在线估计出SPMSM的参数,包括d电感、q轴电感和转子永磁磁势,解决了SPMSM静止状态下的欠秩问题。在此基础上,分析和处理了系统的病态问题,进一步提高了方法的准确性。实验结果验证了方法的有效性。

The paper proposes a method of surface-mounted permanent magnet synchronous motor（ SPMSM） parameter online identification based on adaline neural network（ ANN） with load testing. Firstly,this paper analyzes the rand deficient problem of surface-mounted permanent magnet synchronous motor. Then,through adding of a small perturbation in d-axis current and a small change in rotor speed to aid the estimation of motor parameters（ dq-axis inductances and rotor flux linkage）. And on this basis,analysis and minimization of ill-conditioned problem has been conducted. Finally,the experiment show the effectiveness of the algorithm.