摘要
针对滚动轴承故障严重程度与复合故障难以准确识别的问题,提出了一个基于提升双树复小波包(Lifting Dual-TreeComplex Wavelet Packet,LDTCWP)和深度小波自编码器(Deep WaveletAuto-Encoder,DWAE)的轴承故障诊断方法。首先,使用迁移学习扩展目标数据量;其次,对轴承振动数据进行3层提升双数复小波包分解,分别计算各子频带信号的样本熵、排列熵和能量矩,作为初始特征向量;最后,将初始特征向量输入DWAE,进行二次特征提取并实现故障诊断。实验结果表明,该方法能有效地对滚动轴承进行多种故障类型和多种故障程度的识别,与传统机器学习方法相比,在目标数据较少的情况下也具有较强的泛化能力、特征提取能力和识别能力。
Aiming at the problem that it is difficult to accurately identify the fault severities and compound faults of rolling bearings,a method based on lifting dual-tree complex wavelet packet(LDTCWP)and deep wavelet auto-encoder(DWAE)is proposed.Firstly,the transfer learning strategy is introduced to extend the target data amount.Secondly,the vibration data of bearings is decomposed into three layers via lifting dualtree complex wavelet packet.The sample entropy,permutation entropy and energy moment of each sub-band are calculated as raw eigenvectors.Finally,the raw eigenvectors are sent into DWAE for quadratic feature extraction and fault diagnosis.The fault diagnosis experiment results show that the method can effectively identify multiple fault types and multiple fault severities of bearings.Compared with traditional machine learning methods,the proposed method has better generalization ability,feature extraction ability and recognition ability in the case of insufficient target vibration data.
作者
杜小磊
陈志刚
张楠
郭兴国
Du Xiaolei;hen Zhigang;Zhang Nan;Guo Xingguo(School of Mechanical-electronic and Automobile Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Engineering Research Center of Monitoring for Construction Safety,Beijing 100044,China;College of Electrical and Mechanical Engineering,China University of Mining and Technology(Beijing),Beijing 100083,China)
出处
《机械传动》
北大核心
2019年第9期103-108,共6页
Journal of Mechanical Transmission
基金
国家自然科学基金(51605022)
北京市属高校基本科研业务费专项资金(X18217)
北京市教育委员会科技计划一般项目(SQKM201710016014)
北京市优秀人才培养资助项目(2013D005017000013)
关键词
滚动轴承
提升双树复小波包
深度小波自编码器
迁移学习
故障诊断
Rolling bearing
Lifting dual-tree complex wavelet packet
Deep wavelet auto-encoder
Transfer learning
Fault diagnosis
