基于时间序列异常检测技术的铁路货车热轴预警  

Early Warning of Freight Train Hotbox Based on Time Series Anomaly Detection

作　　者：刘瑞军 司瑾 史红梅[2] LIU Ruijun;SI Jin;SHI Hongmei(Shenhua Railway Transportation Co.,Ltd.,Beijing 100120,China;School of Mechanical,Electronic and Control Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]神华铁路装备有限责任公司,北京100120 [2]北京交通大学机械与电子控制工程学院,北京100044

出　　处：《铁道机车车辆》2022年第4期24-33,共10页Railway Locomotive & Car

基　　金：国家能源集团重大课题(SHGF-17-56)。

摘　　要：实现铁路货车状态修的其中一个重要内容是对车辆技术状态进行全面掌握。轴承作为车辆的关键零部件,其健康状态对铁路货车的运行安全有很大影响,因此,对轴承健康状态的评价以及对热轴的诊断和预警十分重要。现有的轴承监测系统没有分析故障的发展过程和趋势,缺少热轴的提前预警,因热轴报警造成的停机和扣修会对生产指挥产生很大影响。文中针对神华铁路轨边轴温探测系统积累的热轴样本和轴温数据进行深入分析,建立基于时间序列异常检测的热轴早期多级预警模型,对轴承进行状态监测和热轴预警,提前发现含有安全隐患的轴承,最终实现对轴承检修的针对性指导,提高检修效率,减少热轴发生,从而节省计划外的维护成本。实例验证表明模型可以有效地对热轴进行提前预警。One of the important contents to realize the condition-based maintenance of railway freight cars is to master the running-condition of vehicles.As a key component of vehicle,the failure of railway freight car bearing directly affects operation safety.Therefore,it is of great significance to evaluate bearing condition by monitoring data.The existing bearing monitoring system does not analyze the development process and trend of the fault,and lacks the early warning of the hotbox.The shutdown and buckling caused by the hot shaft alarm will have a great impact on the production command.In this paper,the hotbox samples and bearing temperature data accumulated by Shenhua railway track side bearing temperature detection system are analyzed in depth,and the early multistage early warning model of hotbox based on time series anomaly detection is established.The experimental results show that the model can effectively realize early warning of hotbox.

关 键 词：铁路货车 热轴 状态修 特征工程 异常检测 

分 类 号：U272[机械工程—车辆工程]