复杂多变工况下高铁轮对轴承故障识别研究  

Research on fault identification of high-speed railway wheelset bearing under complex and changeable working conditions

作　　者：杨福瑞 赵嘉健 Yang Furui;Zhao Jiajian(High Speed Railway Engineering College,Shaanxi Railway Institute,Shaanxi Weinan,714000,China;China Railway Construction and Safety Engineering Design Institute,Hebei Shijiazhuang,050000,China)

机构地区：[1]陕西铁路工程职业技术学院高铁工程学院,陕西渭南714000 [2]中铁建安工程设计院有限公司,河北石家庄050000

出　　处：《机械设计与制造工程》2024年第3期92-96,共5页Machine Design and Manufacturing Engineering

基　　金：陕西省教育厅科研专项计划(22JK0328);陕西铁路工程职业技术学院2022年校级科研基金(KY2022-39)。

摘　　要：为提升复杂多变工况下轴承故障识别精度,研究了高铁轮对轴承故障识别方法。复杂多变工况下的轮对轴承振动信号经集合经验模态分解后,获得多个IMF分量及剩余分量,计算各分量的排列熵,以此为初始特征集,将朴素贝叶斯算法和最大均值差异(FSBD)方法相结合筛选有效特征,进行源域深度迁移自编码器训练,将输出参数作为目标域深度迁移自编码器的初始参数,检测目标域特征子集,输出故障识别结果。实验结果表明:该方法可实现轮对轴承振动信号的分解,完成初始特征FSI值的计算,选取特征数为150时,轮对轴承故障识别效果最突出,识别精度达到98.5%。In order to improve the accuracy of bearing fault identification under complex and variable working conditions,a high-speed rail wheelset bearing fault identification method is studied.The vibration signals of wheelset bearings under complex and variable working conditions are decomposed into multiple IMF components and residual components through ensemble empirical mode decomposition.The permutation entropy of each component is calculated as the initial feature set,and the naive Bayesian algorithm and maximum mean difference(FSBD)method are combined to screen effective features.The source domain deep transfer autoencoder is trained,and the output parameters are used as the initial parameters of the target domain deep transfer autoencoder to detect the subset of target domain features and output fault recognition results.The experimental results show that this method can decompose the vibration signal of the wheelset bearing,complete the calculation of the initial feature FSI value,select a feature number of 150,and the fault recognition effect of the wheelset bearing is most prominent with a recognition accuracy of 98.5%.

关 键 词：复杂多变工况 高铁轮对轴承 振动信号 集合经验模态分解 深度迁移 

分 类 号：TH177[机械工程—机械制造及自动化]