基于多特征融合的行星齿轮箱点蚀故障诊断研究  

Research on Pitting Fault Diagnosis of Planetary Gearbox Based on Multi-feature Fusion

作　　者：范志锋 华鉴波 FAN Zhifeng;HUA Jianbo(School of Intelligent Manufacturing,Wuchang Institute of Technology,Wuhan 430065,China)

机构地区：[1]武昌工学院智能制造学院,湖北武汉430065

出　　处：《机械与电子》2022年第9期46-50,共5页Machinery & Electronics

基　　金：湖北省教育厅科学研究计划项目(B2021328);武昌工学院校级科研创新团队项目(2019T01)。

摘　　要：为了识别行星齿轮箱的齿面点蚀故障,通过刚柔耦合仿真获得健康和3种不同点蚀程度行星齿轮箱的箱体振动信号。对获得的4种状态的箱体振动信号进行变分模态分解后,计算每个本征模态函数分量的能量值、峭度因子和信息熵,基于能量值、峭度因子和信息熵多特征融合构建高维特征向量,采用支持向量机分类器对4种状态的行星齿轮箱进行识别。结果表明,基于变分模态分解的本征模态函数分量的能量值、峭度因子和信息熵构建的15维特征向量,采用支持向量机分类器能够准确识别健康和3种不同点蚀程度齿轮的类型。In order to identify the pitting fault of planetary gearbox,the vibration signals of healthy planetary gearbox and three different pitting degrees are obtained by rigid-flexible coupling simulation.The energy value,Kurtosis factor and information entropy of each Intrinsic Modal Function component are calculated after the variational mode decomposition of the four state vibration signals.Based on the fusion of energy value,Kurtosis factor and information entropy,a high dimensional eigenvector is constructed,and a Support Vector Machine classifier is used to identify the four states of planetary gearbox.The results show that based on the 15-dimensional eigenvector constructed by the energy value,Kurtosis factor and information entropy of the Intrinsic Modal Function component based on the variational mode decomposition,the Support Vector Machine classifier can accurately identify the types of healthy and three kinds of gear with different pitting degree.

关 键 词：行星齿轮箱 点蚀 变分模态分解 多特征融合 故障诊断 

分 类 号：TH132.4[机械工程—机械制造及自动化] TP277[自动化与计算机技术—检测技术与自动化装置]