基于自适应加权多尺度组合形态滤波的轴承故障特征提取研究  

Feature extraction of bearing faults based on adaptive weighted multi-scale combination morphological filtering

作　　者：韩笑乐[1] 胡天中 余建波[1] HAN Xiaole;HU Tianzhong;YU Jianbo(School of Mechanical Engineering,Tongji University,Shanghai 201804,China)

机构地区：[1]同济大学机械与能源工程学院

出　　处：《振动与冲击》2020年第1期245-252,共8页Journal of Vibration and Shock

基　　金：国家自然科学基金(51375290;71777173);中央高校基本科研业务费项目;上海科委创新科技行动计划(17511109204)资助项目

摘　　要：针对滚动轴承振动信号在进行早期故障诊断时往往会伴随着噪声干扰的问题,提出了一种选择性自适应加权多尺度组合形态滤波(Adaptive Weighted Multi-scale Combination Morphological Filtering,AWMCMF)方法,从振动信号中提取故障特征。首先,采用三种组合算子构成一组新的形态算子,可有效地提取出信号中的正负冲击特征;其次,基于新算子提出了加权多尺度形态滤波方法,并将Teager能量峭度作为评判指标为各尺度提供优化的权值;最后,将选择性自适应权值与多尺度算子进行加权绑定得到优化的故障特征提取结果。通过仿真信号和轴承故障振动信号的结果表明,该方法能有效地滤除噪声并提取故障特征。Aiming at the problem of vibration signals of rolling bearings being often accompanied by noise interference during early fault diagnosis,a selective adaptive weighted multi-scale combination morphological filtering(AWMCMF)method was proposed to extract fault features from vibration signals.Firstly,3 types of combined operators were used to form a group of new morphological operators being able to effectively extract positive and negative impact features in vibration signals.Secondly,based on new operators,a weighted multi-scale morphological filtering method was proposed,and Teager energy kurtosis was taken as the evaluation index to provide optimized weights for various scales.Finally,the weighted binding was performed for selective adaptive weights and multi-scale operators to obtain the optimized fault feature extraction results.The results of simulated signals and bearing faulty vibration signals showed that the proposed method can be used to effectively filter noise and extract fault features.

关 键 词：滚动轴承 故障诊断 振动信号 冲击特征 多尺度形态滤波 

分 类 号：TH165.3[机械工程—机械制造及自动化] TN911.7[电子电信—通信与信息系统]