基于VMD和MSSST的密集化增强轴承故障特征提取  

Feature Extraction of Bearings Fault Based on Densification Enhancement Based on VMD and MSSST

作　　者：薛亚晨 郑小霞 XUE Yachen;ZHENG Xiaoxia(School of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China)

机构地区：[1]上海电力大学自动化工程学院,上海200090

出　　处：《上海电力大学学报》2025年第2期190-197,共8页Journal of Shanghai University of Electric Power

基　　金：国家电网有限公司科技项目(52094023003R);教育部海上风电技术工程研究中心项目。

摘　　要：由于轴承信号具有非线性和非平稳性特点,因此传统的轴承故障特征提取方法在特征选取和噪声处理方面存在困难。为解决上述问题,提出了一种基于变分模态分解(VMD)和多重同步压缩S变换(MSSST)的密集化增强轴承故障特征提取方法。该方法使用VMD算法对轴承振动信号进行分解并筛选有效的本征模态函数(IMF)分量进行信号重构,通过MSSST算法将重构信号转化为高分辨率时频图,进而提取得到高密集度的时频特征。实验结果表明,各种时频特征提取方法中,所提的VMD与MSSST相结合的密集化增强算法的时频分辨率最高,Rényi熵值最低,相比于连续小波变换(CWT)、短时傅里叶变换(STFT)、S变换(ST)和MSSST算法,该算法的Rényi熵值分别降低了53.37%、37.02%、44.23%和8.34%,展现出较好的特征提取能力。For the problems of the nonlinearity and non-stationarity of bearing signals,traditional bearing fault feature extraction methods have difficulties in feature selection and noise processing.To solve the above mentioned problems,a bearing feature extraction method based on densification enhancement using variational mode decomposition(VMD)and multi-synchronous squeezing S transform(MSSST)is proposed.This method uses VMD to decompose the bearing vibration signal and screen the effective intrinsic mode functioncomponents(IMF)for signal reconstruction.The reconstructed signal is converted into a high-resolution time-frequency diagram through MSSST,and then the highly dense time-frequency features are extracted.The experimental results show that,among all extraction methods the proposed feature extraction method using VMD and MSSST has the lowest Rényi entropy value and the highest time-frequency resolution.Compared to continuous wavelet transform(CWT),short-time Fourier transform(STFT),S transform(ST),and MSSST,the proposed algorithm reduces Rényi entropy values by 53.37%,37.02%,44.23%,and 8.34%,respectively,demonstrating a strong feature extraction capability.

关 键 词：轴承故障 特征提取 密集化增强 变分模态分解 多重同步压缩S变换 

分 类 号：TH133.33[机械工程—机械制造及自动化]