基于BP神经网络的感应电动机转子故障诊断方法  

Rotor Fault Diagnosis for Induction Motor Based on BP Neural Network

作　　者：张晓春[1] 杜卫东[1] 许允之[1] 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221116

出　　处：《煤矿机电》2014年第3期89-91,共3页Colliery Mechanical & Electrical Technology

摘　　要：分析定子电流频谱特性诊断感应电动机转子故障,为了避免故障特征的边频分量被基频"旁瓣"淹没,利用扩展Park矢量法提取故障特征。在此基础之上,对所得到的故障特征矢量进行归一化,作为BP神经网络的输入量,同时以故障类型作为BP神经网络的输出量。设计BP神经网络结构,利用输入输出量对BP神经网络进行训练,进而实现感应电动机转子故障的自动识别。In order to avoid the inundation of the edge frequency components of the fault characteristics by the lobes of basic frequency, while the stator current spectrum characteristics are analyzed for rotor fault diagnosis of the induction motor, the extended Park vector method is adopted to extract the fault characteristics. On this basis, the structure of BP neural network is designed, taking the normalized fault characteristic vectors as its input and the fault style as its output. The network is trained by these input and output to realize the auto-identification of the induction motor＇ s rotor fault.

关 键 词：电流频谱分析 PARK矢量 BP神经网络 故障诊断 

分 类 号：TM346.2[电气工程—电机]