摘要
动车组转向架轴箱的寿命作为衡量转向架性能的重要指标,主要受材料、工艺、质量、载荷、保养、工况等因素影响。为解决单一工况预测轴箱故障发生时间不准确问题,需充分考虑多工况因素,基于全生命周期构建转向架轴箱剩余寿命预测模型。本文通过对比分析多工况与轴箱的相互影响关系,采用大数据和机器学习算法,设计出一种基于长短记忆神经网络(LSTM)的轴箱相对温升与里程的剩余寿命预测方法。该方法能精确地刻画轴箱性能退化特征模型,可在动车运行过程中实时预测转向架轴箱故障发生率,较大幅度地提高动车组转向架轴箱剩余寿命预测的实效性、准确性。
As an important index to measure the performance of bogie, the life of bogie axle box for multiple unit train is mainly affected by material, process, quality, load, maintenance, working condition and so on. In order to solve the problem of inaccurate fault occurrence time of axle box prediction under single working condition, it is necessary to fully consider the factors of multiple working conditions and construct the remaining life prediction model of bogie axle box based on the whole life cycle. In this paper, by comparing and analyzing the interaction between multi-working conditions and axle boxes, a residual life prediction method based on long-short memory neural network(LSTM) for relative temperature rise and mileage of axle boxes is designed by using big data and machine learning algorithm. This method can accurately depict the characteristic model of axle box performance degradation, predict the fault incidence of bogie axle box in real time during the operation of high-speed train, and greatly improve the effectiveness and accuracy of the residual life prediction of bogie axle box for multiple unit train.
作者
赵珂
顾佳
姜喜民
ZHAO Ke;GU Jia;JIANG Xi-min(City College,Kunming University of Science and Technology,Kunming 650051,China;China Railway Rolling Stock Corporation Qingdao Sifang Co.LTD,Qingdao 266111,China)
出处
《软件》
2020年第3期219-224,共6页
Software
关键词
铁路运输
轴箱剩余寿命预测
长短记忆神经网络(LSTM)
动车组
动车组转向架
故障预测
Railway transportation
Prediction of residual life of axle box
Long short term memory network (LSTM)
CRH train
Bogies for multiple unit train
Failure prediction
