基于全矢包络融合双层降噪处理的轴承故障特征提取  

Bearing Fault Feature Extraction Based on Full-vector Envelope Fusion and Double-layer Denoising

作　　者：瞿红春[1] 周大鹏 贾柏谊 郑剑青 QU Hongchun;ZHOU Dapeng;JIA Baiyi;ZHENG Jianqing(College of Aeronautical Engineering,Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学航空工程学院,天津300300

出　　处：《噪声与振动控制》2023年第1期135-140,184,共7页Noise and Vibration Control

摘　　要：针对轴承故障信号受背景噪声影响,而难以准确提取故障冲击特征的问题,提出一种噪声辅助多元经验模态分解(Noise-assisted Multivariate Empirical Mode Decomposition,NA-MEMD)与全矢包络快速独立分量分析(Fast Independent Component Analysis,FastICA)相结合的轴承故障特征提取方法。该方法将同源双通道信号进行NAMEMD分解,根据相关性系数选取包含故障特征的固有模态函数(Intrinsic Mode Function,IMF)进行重构;对重构信号进行快速独立分量分析,最后进行全矢包络融合,提取轴承故障特征。对实际轴承信号的分析验证该方法能有效提取完整高阶故障频率,同时降低包络谱特征统计参数的冗余。Aiming at the problem that fault impact features of fault bearings are difficult to be accurately extracted due to the influence of background noise, a fault feature extraction method of bearings was proposed based on the combination of noise-assisted multivariate empirical mode decomposition(NA-MEMD) and full vector envelope Fast independent component analysis(FastICA). In this method, the homologous dual-channel signals were decomposed by NA-MEMD, and the IMF components containing fault features were selected for reconstruction according to correlation coefficients. FastICA was performed on the reconstructed signals. Finally, full-vector envelope fusion was performed to extract bearing fault features. Through the verification of actual bearing signals, it is proved that the proposed method can effectively extract complete high-order fault frequencies and reduce the redundancy of envelop spectrum characteristic statistical parameters.

关 键 词：故障诊断 噪声辅助多元经验模态分解 快速独立分量分析 全矢包络谱 特征提取 

分 类 号：TH133.3[机械工程—机械制造及自动化]