基于小波包分析和有向无环图支持向量机的电机故障诊断研究  

Study on Motor Fault Diagnosis Based on Wavelet Packet Analysis and Directed Acyclic Graph Support Vector Machine

作　　者：崔晨[1] 雷晓犇[1] 张曜晖[1] 范炳奎[1] 

机构地区：[1]空军工程大学工程学院,西安710038

出　　处：《煤矿机械》2009年第10期235-238,共4页Coal Mine Machinery

摘　　要：针对电机振动信号的特点,提出一种基于小波包分析和有向无环图支持向量机的故障诊断方法,将电机不同故障下的振动信号进行小波包分解与重构,提取频带能量作为特征向量,应用有向无环图支持向量机建立从特征向量到故障模式之间的映射,实现对电机的故障诊断。结果表明,该方法能准确有效地诊断电机故障。Aiming at the characteristics of motor vibration signal, a fault diagnosis method is put for- ward based on wavelet packet analysis and directed acyclic graph support vector machine. Motor vibra- tion signals under different fault conditions is decomposed and reconstructed with the theory wavelet packet, the frequency branch energy are picked out as eigenvector. Then the mapping relation from eigenvector to fault modes is established with directed acyclic graph support vector machine that the motor fault analysis is achieved. As proved in the simulation, the method is effective to diagnose motor fault accurately.

关 键 词：电机 小波包分析 有向无环图支持向量机 故障诊断 

分 类 号：TM32[电气工程—电机] TP206[自动化与计算机技术—检测技术与自动化装置]