Analysis Of Accident Violation Behaviors In Railway Transport Operation Based On HFACS-Bayesian Network

Safety production is the eternal theme of railway transportation enterprises.In-depth analysis of accidents through scientific theories and methods,accurate identification of key risks in accidents,and targeted control are essential to effectively reduce accidents.Based on the HFACS and Bayesian network,this paper studied the factors of illegal behavior in railway transport operation accidents.First,on the basis of the HFACS model,a system for classifying and analyzing railway transport operation accidents was constructed,and the causal factors of accidents were divided into four levels:management factors,prerequisites for illegal behavior,illegal behavior,and accidents.Factors in 111 accident cases that occurred from 2018 to 2022 were extracted and integrated according to the four-level classification.Then,based on the hierarchical classification structure of the improved HFACS model,the chi-square test method was used for structural learning to construct a Bayesian network model for illegal behavior in railway transport operation accidents.The maximum likelihood estimation method was used to solve the network parameters and form the conditional probability table of each node in the Bayesian network.The combination of HFACS and Bayesian network provides a novel modeling approach for railway transport operation accident research.Finally,based on the constructed Bayesian network model,Ge NIe software was used for quantitative analysis of illegal behavior factors in railway transport operation accidents,identifying key and sensitive factors of accidents,and analyzing case studies to determine the key control points for illegal behavior factors.The results showed that vehicle inspection,plan preparation,communication review,anti-slip device removal,and derailment confirmation were among the 13 types of illegal behaviors that should be controlled by vehicle departments.In response to the risks of key illegal behaviors,this paper proposes corresponding control measures from two perspectives: overall control strategies and "three prevention" control measures,which can serve as references for practitioners.