基于VMD-LPP融合的轴承故障特征频率提取  

Bearing Fault Frequency Extracting Method Based on Combination of LCD and LPP

作　　者：刘勇[1] 张楠楠[2] LIU Yong;ZHANG Nannan(Shool of Mechatronics and New Energy Automobile,Yichun Vocational and Technical College,Yichun Jiangxi,336000,China;Shool of Material Science and Engineering,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]宜春职业技术学院机电与新能源汽车学院,江西宜春336000 [2]沈阳工业大学材料科学与工程学院,沈阳110870

出　　处：《机械设计与研究》2021年第3期89-93,共5页Machine Design And Research

基　　金：辽宁省教育厅科学研究项目(LJGD2019002)。

摘　　要：为提高轴承故障特征频率的提取效果,提出了变分模态分解(VMD)和局部保持投影(LPP)相融合的轴承故障特征频率提取方法。该方法主要有三个步骤:一是利用VMD对信号进行分解,得到若干个本征模态分量(IMF),并将各分量组成高维信号矩阵;二是利用LPP对高维信号矩阵进行降维得到低维信号矩阵,而后进行信号重构,得到重构信号;三是对重构信号进行包络分析,根据包络谱中突出的频率成分判断轴承故障类型。轴承故障诊断实例验证了方法的有效性。In order extracting bearing fault frequency effectively,a fault frequency extracting method which have three steps of motor bearing based on variational modal decomposition(VMD)and locality preserving projections(LPP)is introduced in this paper.Firstly,bearing original fault signals are decomposed into intrinsic mode component(IMF)through VMD and high dimensions signal matrix which composed by IMF components.Secondly,the LPP was used to reduce the dimension of the high dimensions matrix and the low dimensions matrix can been obtained,thus can obtain reconstructed signal.Thirdly,the fault frequency can be identified by the envelope spectrum which obtained by envelope analysis.The superiority of this method is verified by bearing fault diagnosis living example.

关 键 词：变分模态分解 局部保持投影 包络分析 轴承 故障特征频率 

分 类 号：TH165.3[机械工程—机械制造及自动化]