基于自适应层次聚类的无绝缘轨道电路补偿电容故障诊断方法

针对目前无绝缘轨道电路JTC(Jointless Track Circuit)补偿电容故障诊断方法对补偿电容多故障情况检测不准确、无法充分利用历史监测数据且算法及时性不高、过度依赖于模型的情况,提出基于自适应层次聚类的JTC补偿电容故障诊断方法。通过定义相应类内类间距离判据,提出自适应层次聚类的算法流程。根据机车信号远程监测系统获取的历史监测数据,结合补偿电容故障的影响特征和规律,建立历史数据特征矩阵与样本矩阵,并提出相应的补偿电容多故障诊断策略。实验表明,本算法可以正确、有效地对补偿电容故障情况进行检测,能够提高历史监测数据的利用率。

A Fault Diagnosis Method for Compensation Capacitor of JTC Based on Self-adaptive Hierarchical Cluster

In view of the fact that the present fault diagnosis methods of JTC compensation capacitor are not accurate for compensation capacitor multi-fault detection, the historical monitoring data are not fully used with low algorithm timeliness and over-reliance on models, this paper presents a new fault diagnosis method based on self-adaptive hierarchical clustering. By defining a criterion of in-class and between-class distance, the algorithm of self-adaptive hierarchical clustering is presented. According to the historical monitoring data from cab signal remote monitoring system, the historical data characteristic matrix and sample matrix are established on the basis of the influence of compensation capacitor's fault, and then the multiple compensation capacitor fault diagnosis strategy is put forward. Experiments show that this algorithm can detect the fault of compensation capacitor correctly and effectively, and make full use of historical monitoring data.