HFACS-RAs Based Railway Accident Casual Factor Modeling And Hybrid Learning Approach

As a complex social system,the improvement of the technology for the railway system has increased the reliability of the equipment,while the accompanying high integration and automation degree made the human-machine interaction more and more difficult.Thus,the human and organizational factors have became the main contributing factors for most railway accidents.Given that the accident reports in text forms and the uncertainty of the human and organizational factors bring difficulty to the railway accident investigation,a five-phase hybrid learning approach for the railway accidents in China has been designed in this study.This approach was applied to conduct the qualitative and quantitative analysis for the involved human and organizational factors,and some advices on the planning,implementation and evaluation of the intervention measures were given.The main works of this study are as follows:(1)The Human Factor Analysis and Classification System-Railway Accidents(HFACS-RAs)framework was designed based on the historical accident data,which was used to conduct the classification for the human and organizational factors.Not only the errors of shop floor personnel can be identified by HFACS-RAs,its brief hierarchical framework consisting of Unsafe Acts,Preconditions of Unsafe Acts,Unsafe Supervision and Organizational Influeces also made in-depth exploration on the gonverming variables which affect human performances and led to accident indirectly.Thus,the HFACS-RAs framework provides a valid classification criterion to the disorganized human and organizational factors.(2)Considering the hierarchical framework of the HFACS-RAs framework,the quantitative influence relationship analysis method based on the Analytic Network Process(ANP)method was designed.The ANP method divided the influence relationships of the classified factors into inner dependencies and outer dependencies respectively,and the relative weights of relevant factors were computed by the Supermatrix.The inner dependencies were analyzed using the Fuzzy Decision Making Trail and Evaluation Laboratory(F-DEMATEL)method,so the unreliability caused by the expert judgement of the parameterization during the ANP process can be crippled.The key human and organizational factors can be recognized by the causal diagram and relative weights,so the scope of the defects of the organization can be reduced.(3)The proposed method was applied to analyze the 7.23 Yong-Wen line major transportation accident.The identified human and organizational factors were classified by HFACS-RAs and corresponding weights were calclulated using the ANP and DEMATEL method.Finally,the main accident casual factors are identified and the proposed method can be demonstrated.