基于文本信息抽取的列控车载设备故障发现

Fault discovery based on text information extraction for on-board equipment of CTCS

列车运行控制系统车载安全计算机日志文件和地面设备维护工作人员撰写的交接班记录——质量分析工区工作记录单,包含大量反映车载设备运行状态的信息,是车载设备故障发现和诊断的重要依据。对这两类文本数据的利用目前还局限于工作人员人工记录、查询和解读,存在效率低、主观性强、易遗漏出错等问题。针对以上问题,以车载安全计算机日志文件和质量分析工区工作记录单为研究对象,设计信息抽取方法及故障发现自动化方案,以代替现有人工信息抽取和故障记录的工作方式。分析了车载安全计算机日志文件的数据特点,利用分词、智能信息抽取算法对非结构化的车载设备数据进行处理,并结合历史质量分析工区工作记录单数据进行故障匹配,从而构建故障分析字典,实现基于正则表达式算法的故障语句自动定位。设计实现了故障自动发现软件,验证了故障发现方案的可行性和软件的功能。

The on-board safety computer of Chinese Train Control System generates a large number of Log files during operation every day,which together with the quality analysis record sheet,record the running status of all equipment and provide data support for the fault discovery and processing of the equipment of train control system.The use of these two types of text data is still limited to manual record,query and analysis,which has the problems of low efficiency,strong subjectivity and easy omission and error.The on-board safety computer Log files and the quality analysis record sheet were taken as the research object,and designs an information extraction method and the fault discovery automation scheme to replace the existing manual working mode.The information of Log files and manual recording by the staffs were analyzed,the word segmentation algorithm and information extraction algorithm were selected to find useful infor-mation automatically from two kinds of record to avoiding the cumbersome manual search.Then a failure analysis dic-tionary was built.The automatic fault discovery application was established by using regular expression algorithm.The automatic fault discovery and analysis functions of the application were validated by experiment.