基于TPE-Informer的列车风源系统故障预警研究  

Research on fault prognosis for air supply system based on TPE-Informer

作　　者：翟鸿儒 姚爱琴[1] 孙运强[1] 赵文强 石喜玲 Zhai Hongru;Yao Aiqin;Sun Yunqiang;Zhao Wenqiang;Shi Xiling(School of Information and Communication Engineering,North University of China,Taiyuan 030051,China;School of Electrical and Control Engineering,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学信息与通信工程学院,太原030051 [2]中北大学电气与控制工程学院,太原030051

出　　处：《国外电子测量技术》2024年第9期111-120,共10页Foreign Electronic Measurement Technology

基　　金：青岛市科技成果转化专项园区培育计划(2311400010HX);山西省自然科学基金(20210302123062)项目资助。

摘　　要：针对轨道交通列车的风源系统在每日高强度和负荷的运行环境下容易发生故障的问题,结合空气压缩机作为系统的核心部件受多种因素影响呈间歇性运行进而导致监测数据在时间维度上分布不均,现有基于连续运行数据的故障预警方法难以适用的情况,研究了一种适应列车风源系统的故障预警框架。首先结合不同类型数据的特点提取多维时序特征,并使用Informer模型学习正常数据特征,同时应用基于树结构的贝叶斯优化算法(TPE)优化模型参数。然后通过分析模型预测值与实际测量值的残差,提出了一种预警指标计算方法。在公开的列车风源故障数据集MetroPT3上的仿真实验结果表明,该预警框架最早于故障前220个周期发出预警,且最晚于前22个周期发出预警,并对误报具有良好的鲁棒性。Addressing the susceptibility of rail transit air supply system to faults under high-intensity and high-load daily operations,this study presents a fault warning framework tailored to the intermittent operation of the core component,the air compressor.The framework is designed to handle the uneven temporal distribution of monitoring data due to various influencing factors.Multidimensional temporal features are extracted from different types of data,and the Informer model is utilized to learn the characteristics of normal operation,with the TPE algorithm optimizing model parameters.A method for calculating a warning index is proposed based on the analysis of residuals between predicted and actual measurements.Simulation experiments on the MetroPT3 Air Supply fault dataset confirm the framework's can issue warning signals as early as 220 cycles prior to a fault and as late as 22 cycles before the fault occurs,and exhibiting its robustness against false alarms.

关 键 词：风源系统 空气压缩机 深度学习 故障预警 

分 类 号：TH45[机械工程—机械制造及自动化] TN06[电子电信—物理电子学]