基于DFD-DBSCAN的高速列车电池组多故障诊断方法  

Multi-fault diagnosis method of high-speed train battery packs base on DFD-DBSCAN

作　　者：向超群[1] 席振 左明洁 毕福亮 成庶[1] 于天剑[1] XIANG Chaoqun;XI Zhen;ZUO Mingjie;BI Fuliang;CHENG Shu;YU Tianjian(School of Transportation Engineering,Central South University,Changsha 410075,China;CRRC Changchun Railway Vehicles Co.,Ltd.,Changchun 130062,China)

机构地区：[1]中南大学交通运输学院,湖南长沙410075 [2]中车长春轨道客车股份有限公司,吉林长春130062

出　　处：《铁道科学与工程学报》2024年第7期2980-2988,共9页Journal of Railway Science and Engineering

基　　金：国家自然科学基金资助项目(52072414)。

摘　　要：高速列车电池作为备用电源,被广泛应用于辅助供电系统以维持高速列车控制系统的正常运转,其可靠性涉及行车安全。列车频繁起停、频繁加减速以及震动等多种复杂运行环境易导致电池单体故障和连接故障。为了保证高速列车的安全运行,高速列车电池组的状态检测与多故障诊断研究备受关注。目前,针对高速列车电池组的多故障诊断方法的研究尚属空白,提出一种基于改进离散弗雷歇距离(Discrete Fréchet Distance, DFD)和自适应密度聚类(Density-Based Spatial Clustering of Applications with Noise, DBSCAN)的高速列车电池组的实时多故障诊断方法,以准确识别电池组的连接故障和单体故障。以高速列车电池作为研究对象,通过设计适用于高速列车电池组的电压交叉测量方法,使得电池电压和连接板电压与不同的电压传感器相关联,并通过DFD算法对电池组的故障特征进行提取,将电压偏移率与DFD共同作为故障诊断模型的参数输入以提高算法的鲁棒性与可靠性,接着引入DBSCAN算法自动对故障诊断并定位。为了保证算法的实时性,利用基于滑动窗口的遗忘机制实时地对采样数据进行诊断。通过实验对所提出的方法进行验证,结果表明该方法可及时有效地诊断电池组的单体故障与连接故障并准确定位,弥补了高速列车电池组多故障诊断方法研究的缺失,对提高轨道列车的行车安全具有工程实用意义。The battery system in high-speed trains,which serves as a backup power source,is extensively utilized in auxiliary power supply systems,with its reliability being crucial for driving safety.The complex operational environment of trains,characterized by frequent starts and stops,acceleration and deceleration,as well as vibrations,can easily cause failures in individual battery cells and connections.To ensure the safe operation of high-speed trains,the state monitoring and multi-fault diagnosis of high-speed train battery packs have garnered significant attention.Currently,there is a gap in the research on multi-fault diagnostic methods for high-speed train battery packs.This paper proposed a real-time multi-fault diagnostic method for high-speed train battery packs,based on an improved Discrete Fréchet Distance(DFD)and an adaptive Density-Based Spatial Clustering of Applications with Noise(DBSCAN)algorithm.This method aimed to accurately identify connection and individual cell faults in the battery pack.Focusing on high-speed train batteries,this study designed a voltage cross-measurement method suitable for high-speed train battery packs,linking battery voltage and connection board voltage with different voltage sensors.The DFD algorithm was employed to extract fault characteristics of the battery pack.The voltage deviation rate and DFD were used together as input parameters for the fault diagnosis model to enhance the algorithm's robustness and reliability.Subsequently,the DBSCAN algorithm was introduced for automatic fault diagnosis and localization.To ensure the real-time nature of the algorithm,a sliding window-based forgetting mechanism was utilized for real-time diagnosis of the sampled data.Experiments conducted to validate the proposed method demonstrate its effectiveness in timely and accurately diagnosing and pinpointing individual cell and connection faults in battery packs.This approach fills the research gap in multi-fault diagnostic methods for high-speed train battery packs and holds practical eng

关 键 词：高速列车 电池组 故障诊断 弗雷歇距离 DBSCAN算法 

分 类 号：TM912[电气工程—电力电子与电力传动]