基于改进FP-Growth算法和贝叶斯的营业线施工安全风险分析  

Construction safety risk analysis of business lines based on improved FP-Growth algorithm and Bayesian

作　　者：蔡近近 宋瑞[1] 何世伟[1] 赵日鑫 姜俊平 CAI Jinjin;SONG Rui;HE Shiwei;ZHAO Rixin;JIANG Junping(Key Laboratory of Transport Industry of Big Data Application Technologies for Comprehensive Transport,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学综合交通运输大数据应用技术交通运输行业重点实验室,北京100044

出　　处：《铁道科学与工程学报》2024年第8期3370-3381,共12页Journal of Railway Science and Engineering

基　　金：国家自然科学基金资助项目(62076023);中国国家铁路集团有限公司科技研究开发计划课题资助项目(K2021X002)。

摘　　要：铁路营业线施工事故致因因素众多且存在关联关系,挖掘事故致因之间的关联关系和因果关系对事故的预防管控具有重要意义。通过文本挖掘对2010—2022年某路局营业线事故调查报告进行处理,提取出51个事故致因因素。基于事故因果连锁理论将致因因素分为人因层、设备层、环境层、管理层4个层级进行分层分析,构建铁路营业线施工事故致因体系。基于压缩算法和差分编码对传统FP-Growth算法进行改进,以此对铁路营业线施工事故致因进行挖掘,找到满足提升度要求的高支持度关联规则和高置信度关联规则,发现关键致因关联和事故致因规律。基于贝叶斯网络理论、致因关联关系和专家经验建立营业线施工安全风险贝叶斯网络,结合复杂网络理论分析网络节点度、聚类系数与节点介数等特征,找到关键致因因素。在此基础上,运用因果推理和故障诊断推理进一步剖析营业线施工过程中的高风险致因,并从“人防、物防、技防”3方面提出预防管控措施。案例结果表明:施工人员操作不当、施工造成接触网故障、施工导致设施设备侵限、施工作业损害电缆设备、施工造成轨道电路故障和施工、检修、清扫设备耽误列车类事故之间的关联关系较为频繁,且为红光带事故的高概率致因,在施工作业过程中应多层次重点预防管控。研究成果为铁路营业线施工安全管理提供一种新的风险分析方法。Many factors cause the construction accidents of railway business lines and are inter-correlated.It is important to explore the correlation and causation between the causes of accidents for the prevention and control of accidents.The investigation report of a business line accident in a railway bureau from 2010 to 2022 was processed by text mining,and 51 accident cause factors were extracted.The cause factors were divided into four levels:human cause level,equipment level,environment level,and management level for hierarchical analysis based on the causal chain theory of accidents,and the cause system of railway business line construction accidents was constructed.The traditional FP-Growth algorithm was improved to mine the causation of railway business line construction accidents based on compression algorithm and differential coding,find high-support association rules and high-confidence association rules that satisfy the requirements of enhancement degree,and find the key causation association and accident causation law.Bayesian network for construction safety risk of business lines was established based on the Bayesian network theory,causative correlation relationship,and expert experiences.Combined with complex network theory,the characteristics of network node degree,clustering coefficient,and node internode number were analyzed to find the key causative factors.On this basis,the causal reasoning and fault diagnosis reasoning were used to further analyze the high-risk causes in the construction process of business lines,and the prevention and control measures were proposed from the three aspects of“civil defense,physical defense,and technical defense”.The results of the case show that:improper operation of construction personnel,failure of overhead contact system caused by construction,encroachment of facilities and equipment caused by construction,damage to cable equipment caused by construction,track circuit failure caused by construction,and maintenance and cleaning equipment delay of train accidents

关 键 词：铁路营业线施工事故 改进FP-Growth算法 关联规则挖掘 贝叶斯网络推理 致因体系 

分 类 号：U298.5[交通运输工程—交通运输规划与管理]