基于主成分分析和神经网络的电动机故障诊断方法研究  

Research on Fault Diagnosis Method of Motor Based on PCA and Neural Network

作　　者：许允之[1] 许智颖 方磊[1] 郭西进[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 [2]大连市旅顺口区第57中学,辽宁大连116041

出　　处：《煤矿机电》2016年第5期27-32,共6页Colliery Mechanical & Electrical Technology

摘　　要：为能有效地对电动机转子断条故障初期进行检测,以小波包分析方法为基础,根据能量频谱中能量值的变化来诊断电动机故障。根据能量值的大小来判别电动机故障的严重程度,再利用小波包结合主成分分析(PCA)提取能反映原始数据集的特征向量,最后利用径向基函数(RBF)神经网络来对电动机故障类型进行识别。仿真验证了该方法能有效地提取电动机的故障信号量,并能正确地识别电动机的故障类型。In order to effectively test the preliminary fault of motor broken rotor bar, diagnoses the motor faults according to the change of energy value in the energy spectrum, which is based on wavelet package analysis method, or judges the severity of the motor fault according to the size of energy value. Then uses the wavelet packet in combination with the principal component analysis （PCA） to extract the characteristic vector that reflecting the original data set. Finally, uses the radial basis function （RBF） neural network to identify the fault type of the motor. The simulation verifies that the induction motor can effectively extract the fault signal vector of the motor, and can correctly identifies the fault type of the motor.

关 键 词：电动机故障诊断 频谱分析 小波包分析 主成分分析(PCA) 

分 类 号：TM343.2[电气工程—电机] TP206.3[自动化与计算机技术—检测技术与自动化装置]