基于地铁运营日志文本挖掘的危险源辨识算法研究  

Research on identification algorithm of hazard sources based on text mining of metro operation log

作　　者：牟庆泉 丁小兵[1] 刘志钢[1] 吴先源 MU Qingquan;DING Xiaobing;LIU Zhigang;WU Xianyuan(School of Urban Railway Transportation,Shanghai University of Engineering Science,Shanghai 200336,China)

机构地区：[1]上海工程技术大学城市轨道交通学院,上海200336

出　　处：《中国安全生产科学技术》2022年第3期204-210,共7页Journal of Safety Science and Technology

基　　金：2020年上海高校市级重点课程建设项目(S202010001)。

摘　　要：为探究地铁运营过程中致使险性事件发生的关键因素,识别关键危险源,加强对危险事件的预防管控,研究基于地铁运营日志文本挖掘的地铁运营危险源辨识算法,通过引入自定义专业术语词典和自定义停用词典,删除干扰数据并进行数据分词,实现对地铁运营日志的预处理,进一步构建进行危险源辨识的AFP-tree算法,实现对日志中关键危险源的挖掘。研究结果表明:AFP-tree算法能够显著提升计算效率,通过分析共挖掘出25类有效关键危险源,研究结果可为地铁运营单位实现“事前”的事故预防提供重要依据。In order to explore the main causes of key hazards in the process of metro operation,identify key hazard sources,and effectively strengthen the prevention and control of dangerous events,a hazard identification algorithm of metro operation based on log text mining was designed.Through introducing into the customized terminology dictionary and the customized stop words dictionary,the interference data was deleted and the data was segmented to realize the preprocessing of the metro operation log.Then,the AFP-tree algorithm of hazard sources identification was constructed to realize the mining of key hazard sources in the log.The results showed that the AFP-tree algorithm could significantly improve the calculation efficiency.Through analysis,25 types of effective key hazard sources were mined.It can provide important basis for the metro operating company to achieve the“pre-event”accident prevention.

关 键 词：地铁运营安全 危险源辨识 文本挖掘 APRIORI算法改进 关联规则 

分 类 号：X951[环境科学与工程—安全科学]