列车轴箱轴承在途鲁棒可视化故障诊断方法  

Robust and Visual Fault Diagnosis Method for Train Axle-box Bearing During Train Operation

作　　者：付云骁 贾利民[2,3] 杨杰[2] 魏秀琨[2] 秦勇[2,3] FU Yunxiao;JIA Limin;YANG Jie;WEI Xiukun;QIN Yong(CRRC Industrial Institute Co.,Ltd.,Beijing 100070,China;State Key Laboratory of Rail Traffic Control and Safety,Beijing Jiaotong University,Beijing 100044,China;Beijing Engineering Research Center of Urban Traffic Information Intelligent Sensing and Service Technologies, Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]中车工业研究院有限公司,北京100070 [2]北京交通大学轨道交通控制与安全国家重点实验室,北京100044 [3]北京交通大学北京市城市交通信息智能感知与服务工程技术研究中心,北京100044

出　　处：《铁道学报》2018年第12期38-45,共8页Journal of the China Railway Society

基　　金：国家重点研发计划(2016YFB1200402-002);轨道交通控制与安全国家重点实验室自主研究课题(RSC2016ZT006)

摘　　要：为使列车轴箱轴承在非平稳工况下的故障识别更加有效,本文提出基于融合相关熵特征的鲁棒可视化滚动轴承故障诊断方法。通过快速集成经验模态分解FEEMD对轴承振动信号进行时频分解,提取本征模函数IMF矩阵;计算IMF与原始信号的线性相关系LCC作为相关熵的调幅系数,进而通过相关统计计算获得样本集的多维相关熵矩阵CM;利用主元分析PCA对CM进行数据空间变换,通过提取变换后的融合相关熵矩阵ICM,实现相关熵矩阵的可视化。通过实验分别提取匀加速、匀速及匀减速3种运行工况下的滚动轴承ICM特征,通过对比EMD、EEMD和FEEMD 3种信号分解方法,发现FEEMD的信号分解效率更高,且ICM比传统特征对非平稳工况下轴承故障辨识的鲁棒性更好。FEEMD-ICM为轴箱轴承快速、客观且稳定的故障诊断实现提供了可靠的理论依据和技术支持。In order to identify more effectively the faults of train axle box bearings under non-stationary operating conditions,a robust visualization fault diagnosis method for rolling bearing was proposed based on the fusion correlation entropy features.First,the time-frequency decomposition of bearing vibration signals was performed by Fast Ensemble Empirical Mode Decomposition(FEEMD)to extract Intrinsic Mode Function(IMF)Matrix.After that the Linear Correlation Coefficient(LCC)of the IMF and the original signal was calculated as the modulation factor of the correlation entropy.Furthermore,a multi-dimensional correlation entropy matrix(CM)of the sample set was obtained through relevant statistical calculations.Moreover,Principal Component Analysis(PCA)was used to transform the data space of the CM,and the transformed Integrated Correntropy Matrix(ICM)was extracted to realize the visualization of the correlation entropy matrix.Finally,the ICM characteristics of the rolling bearing under three operating conditions of uniform acceleration,uniform speed and uniform deceleration were extracted in the laboratory environment.By comparing the three signal decomposition methods of EMD,EEMD and FEEMD,it has been found that the signal decomposition efficiency of FEEMD is the highest.Simultaneously,the robustness of ICM is proved better than traditional features for bearing fault identification under non-stationary operating conditions.In summary,FEEMD-ICM can provide a reliable theoretical basis and technical support for the rapid,objective and stable fault diagnosis of axle box bearings.

关 键 词：快速经验模态分解 融合相关熵矩阵 主成分分析 滚动轴承 可视化 故障诊断 

分 类 号：U260.3312[机械工程—车辆工程]