Research On Accident Probability Prediction Of Railway Train Service System Based On Bayesian Network Model

Railway is a complex system composed of people,equipment and environment.Accidents will inevitably occur in its operation,and then affect the normal operation of railway.Among these people,the depot system is responsible for the train scheduling and command work,plays a core role in the normal operation of the railway system,is an important part of the study of railway safety management.The operation process of depot system is more complex,and many links have safety risks and hidden accidents.It is an important measure to strengthen the safety management of depot system to identify the key risk points and predict the accident probability effectively by using scientific methods.Bayesian networks solve the posterior probability evaluation method by prior probability and conditional probability related to prior probability,which has obvious advantages in fault prediction and traffic management.Therefore,it is of great significance to improve the level of railway transportation safety management to study the probability prediction of traffic accidents based on Bayesian network model and explore the scientific probability prediction method of railway traffic accidents.First of all,through the investigation of a railway rolling stock system of a railway bureau group,the basic information of railway rolling stock system safety in the past ten years（including operation quantity,accident occurrence and failure type,etc.）is obtained,from which the accident risk factors of a railway bureau group’s depot system are extracted.It provides a basis for the probability prediction of accident occurrence based on Bayesian network model.Then,through the comparison of common risk assessment methods,the interpretation structure model is selected to quantitatively evaluate the accident risk factors of the depot system,and the classification of accident risk grade of the depot system is obtained.Finally,combined with the basic principle of Bayesian network,the accident prediction model of depot system based on Bayesian network is constructed and applied to the accident prediction of depot system of a railway bureau group company,and the accident probability of depot system is obtained,which provides decision-making basis for the fault prevention and control of depot system.Through in-depth analysis of selected research samples,the validity and practicability of the prediction model proposed in this paper are basically verified.It is found that the number of depot system accidents of a selected railway bureau group company fluctuates and increases,and the accidents become more frequent,mainly including personal accidents,receiving and dispatching accidents,shunting accidents,construction accidents and other accidents.The direct risk factors at the upper level include divergent directions,dispatching orders and personal safety;the middle risk factors include emergency treatment,cancellation signal,wrong operation and misoperation,train derailment and wrong shoes;the bottom risk factors include safety management,tail-train operation and train separation.According to the accident prediction of depot system,the accident probability is 36.7%.The risk of emergency treatment,safety management,wrong operation and other aspects is the key to causing the accident of depot system,which requires high attention.