基于改进PSO-VMD的滚动轴承早期故障诊断  

Early Fault Diagnosis of Rolling Bearing Based on Improved PSOVMD

作　　者：何凯 廖玉松[1] 胡斌 黄斯琪 HE Kai;LIAO Yusong;HU Bin;HUANG Siqi(School of Mechanical and Automotive Engineering,Chuzhou Polytechnic,Chuzhou Anhui 239000,China)

机构地区：[1]滁州职业技术学院机械与汽车工程学院,安徽滁州239000

出　　处：《盐城工学院学报（自然科学版）》2023年第2期61-66,共6页Journal of Yancheng Institute of Technology:Natural Science Edition

基　　金：滁州职业技术学院自然科学研究重点项目(YJZ-2020-06)。

摘　　要：针对滚动轴承早期故障信号微弱导致分类识别率低的问题,提出利用复合多尺度模糊熵作为适应度函数的粒子群算法优化变分模态分解,得到多个本征模态分量;利用快速谱峭度图选择最优的本征模态分量,并组成特征向量;将特征向量输入SSA-SVM中进行故障分类。实验结果表明基于复合多尺度模糊熵的PSO-VMD和SSA-SVM的滚动轴承故障诊断更能有效地识别出滚动轴承的早期故障。In response to the problem of low classification recognition rate caused by weak early fault signals of rolling bearings,a particle swarm optimization algorithm using composite multi-scale fuzzy entropy as a fitness function is proposed to optimize variational modal decomposition and obtain multiple intrinsic modal components.The optimal eigenmode components are selected by using the fast spectral kurtosis diagram,and the eigenvectors are formed.The feature vector is input into SSA-SVM for fault classification.The experimental results show that the fault diagnosis of rolling bearings of PSO-VMD and SSA-SVM based on composite multi-scale fuzzy entropy is more effective in identifying early faults of rolling bearings.

关 键 词：复合多尺度模糊熵 粒子群算法 变分模态分解 快速谱峭度图 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]