基于PHM的高速铁路牵引电机滚动轴承健康状态预测  

Health Condition Prediction of Rolling Element Bearings of High-speed Train Traction Motor Based on PHM

作　　者：牛齐明 刘峰[1] 张奕黄[3] NIU Qiming;LIU Feng;ZHANG Yihuang(School of Computing and Information Technology,Beijing Jiaotong University,Beijing 100044,China;Department of Computer Teaching,Hebei University,Baoding 071002,China;School of Electrical Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学计算机与信息技术学院,北京100044 [2]河北大学计算机教学部,河北保定071002 [3]北京交通大学电气工程学院,北京100044

出　　处：《铁道学报》2020年第7期95-101,共7页Journal of the China Railway Society

基　　金：中国铁路总公司科技研究开发计划(2016J007-B)。

摘　　要：针对高速铁路牵引电机滚动轴承健康状态预测问题,提出了基于深度堆叠去噪自编码器累积(DSDAE)和时滞最小二乘支持向量机(TDLSSVM)的预测方法。在提取高速铁路牵引电机滚动轴承健康状态的多种特征后,用深度堆叠去噪自编码器进行特征降维,并累积计算得到相关的健康指标。将健康指标作为训练数据用以构建时滞最小二乘支持向量机(TDSVR)模型,通过对健康指标的预测实现对健康状态进行评估。在公开数据集上做了DSDAE与TDSVR、累积马氏距离(MDCUSUM)与TDSVR、DSDAE与TDLSSVM和MDCUSUM与TDLSSVM四种方案的对比实验;在高速铁路某型号的牵引电机滚动轴承数据集上做了DSDAE与TDLSSVM方案的实验。通过对预测指标的分析可知,DSDAE与TDLSSVM方案可以很好地预测滚动轴承的健康状态变化趋势。Aiming at the health condition prediction of rolling element bearings of high-speed train traction motor,a prediction method based on deep stacking denoising autoencoder(DSDAE)accumulation and time-delay least squares support vector machine(TDLSSVM)was proposed.After the extraction of the features of the health condition of rolling element bearings of high-speed train traction motor,the feature dimension reduction was carried out by DSDAE,and the related health indicators(HI)were obtained by cumulative calculation.The HI was used as training data to build TDLSSVM model,and the health condition was evaluated through the prediction of the HI.A comparative experiment on four schemes of DSDAE and TDSVR,MDCUSUM and TDSVR,DSDAE and TDLSSVM,MDCUSUM and TDLSSVM was done on the open dataset.The experiment on the DSDAE and TDLSSVM scheme was done on the data set of traction motor rolling bearings of a certain type of high-speed train.Through the analysis of the prediction indicators,it can be concluded that the DSDAE and TDLSSVM scheme can well predict the changing trend of the health status of rolling element bearings.

关 键 词：高速铁路装备 故障预测与健康管理 健康指标 状态预测 滚动轴承 

分 类 号：U260.33[机械工程—车辆工程]