基于文本挖掘的高铁信号系统车载设备故障诊断  

Text Mining Based Fault Diagnosis for Vehicle On-board Equipment of High Speed Railway Signal System

作　　者：赵阳[1] 徐田华[2] 

机构地区：[1]中国科学院自动化研究所,北京100190 [2]北京交通大学轨道交通控制与安全国家重点实验室,北京100044

出　　处：《铁道学报》2015年第8期53-59,共7页Journal of the China Railway Society

基　　金：中国铁路总公司科技研究开发计划(2013X015-B);轨道交通控制与安全国家重点实验室自主研究课题(RCS2012ZT005)

摘　　要：本文以故障文本信息为依据,提出基于文本挖掘的高铁信号系统车载设备的故障诊断方法。针对故障追踪表记录的不规范性和随意性,采用主题模型对故障追踪表进行分析和特征提取;在此基础上,考虑到高铁信号系统车载设备故障诊断的不确定性,采用贝叶斯网络作为故障分类的方法。在贝叶斯网络结构的确定中,根据车载设备的特点与领域专家知识,提出适用于车载设备的贝叶斯结构学习算法HDBN_SL。以武广线的现场数据为依据,进行实验分析,测试结果表明本文特征提取以及故障诊断方法具有较好的诊断准确性。Based on fault text data, a fault diagnosis method for vehicle on-board equipment （VOBE） of high speed railway signal system has been proposed. Due to the irregularity and arbitrary nature of fault tracing records, the topic model was used for fault analysis and feature extraction. In addition, considering the uncertainty and complexity of fault diagnosis of VOBE of high speed railway signal system, a Bayesian network （BN） based fault diagnosis system for VOBE was proposed. In the process of deriving an appropriate BN structure for VOBE, subject to the characteristics and domain knowledge of VOBE, a new algorithm called HDBN_ SL （Hierarchical Diagnostic Bayesian Networks-Structure Learning） that is applicable to VOBE was proposed. Experimental analysis was conducted based on the field data from Wuhan-Guangzhou high speed railway signaling systems. The test results showed that the text mining feature extraction and fault diagnosis method delivered better diagnostic accuracy.

关 键 词：故障诊断 高速铁路 车载设备 主体模型 贝叶斯网络 

分 类 号：TP301.2[自动化与计算机技术—计算机系统结构] U238[自动化与计算机技术—计算机科学与技术]