基于统计学与BP神经网络的永磁同步电机故障识别方法

Fault Identification Method of Permanent Magnet Synchronous Motor Based on Statistics and BP Neural Network

摘要: 针对永磁同步电机三相定子不对称故障识别问题，提出了基于参数统计分析提取优质指标与BP神经网络相结合的识别方法。基于永磁同步电机的数学模型建立仿真模型，模拟正常状态和故障状态下的定子电流数据；通过对电流特征指标的统计分析，筛选出时域特征指标中的优质指标；将筛选出来的指标输入BP神经网络进行故障识别，结果表明所提方法可以实现对永磁同步电机定子电阻三相不对称故障的识别，准确率超过96%。

Abstract: To identify the three-phase stator asymmetric fault of permanent magnet synchronous motor (PMSM), a statistical analysis and BP neural network based method is proposed. Firstly, a simula-tion model of PMSM is established to simulate the stator current data in normal state and fault state. Secondly, the time domain feature indexed are extracted by analyzing the current data, and the op-timal indicators are screened out by statistics analysis. Finally, the optimal indicators are used to train a BP neural network to achieve fault identification. The results show that the proposed meth-od can realize the identification of three-phase asymmetric fault of the PMSM, and the accuracy rate is more than 96%.