【摘要】 针对信号集中监测系统(CSM)对25 Hz相敏轨道电路故障诊断能力需要提升的情况，基于传输线理论和LightGBM算法提出一种25 Hz相敏轨道电路的监测方案和故障诊断方法。首先，采用电流监测方式，将轨道电路室内所有可行的电流监测点加入原CSM监测方案中，构成初步监测方案，并利用传输线理论对所有可监测信号进行建模。然后，使用故障注入技术对各常见故障模式下的监测信号进行仿真，构建故障数据集，生成相应故障诊断模型，并基于模型的性能，对初步监测方案进行优化。最后，设计包含诊断阈值与诊断模型再训练过程的故障诊断策略。实验表明，本文方法只需增加对送电端电源屏输出电流的监测，便可实现对25 Hz相敏轨道电路14种常见故障的精确诊断，并可实现对非常见故障的预判、再学习与诊断。此方法可有效提升CSM对25 Hz相敏轨道电路的故障诊断性能，能够满足铁路现场对25 Hz相敏轨道电路的使用和维护要求。更多还原

【Abstract】 To improve the performance of the 25 Hz phase-sensitive track circuit fault diagnosis method based on the centralized signal monitoring system(CSM), an improved monitoring plan and a fault diagnosis method based on the transmission line theory and the LightGBM algorithm were proposed. Firstly, the non-contact current monitoring method was adopted to add all feasible indoor current monitoring points to the original CSM monitoring plan, which constitutes the preliminary monitoring plan proposed in this paper, and the transmission line theory was used to model all the monitorable signals. Secondly, the fault injection technology was used to simulate the monitoring signals in each common fault mode, build the fault dataset, generate the corresponding fault diagnosis model, and optimize the preliminary monitoring plan based on the performance of the model. Finally, this paper designed a fault diagnosis strategy that includes the diagnosis threshold and the model retraining process. Experiments show that the proposed method can accurately diagnose 14 common faults of 25 Hz phase sensitive track circuit only by increasing the monitoring of the output current of the power supply screen at the power transmission end, and can distinguish, relearn and diagnose uncomon faults. The proposed method can effectively improve the fault diagnosis performance of CSM for 25 Hz phase-sensitive track circuits, and can meet the use and maintenance requirements for the circuits.更多还原