Research On Train Real-time Risk Assessment Model Under Relative Moving Block

As large volume transportation,rail transit has undertaken an important job of carrying and transporting passengers and cargo.Since the relative moving block is a necessary way to further improve the capacity of the rail transit as well as realize virtual coupling,its implementation is getting more and more policy support and meets the market’s demands.However,the relative moving block is not an absolutely safe block mode.Moreover,train operation risk assessment faces the challenges of interaction between systems,complex environmental factors,and the coupling relationship between system and environment.Therefore,it is of great significance to clearly monitor the scene parameters of safety boundary,dynamically infer the operation risks through the collected scene data,and extend the monitoring of train safety risks and tracing distance to provide decision support for drivers,avoiding accidents and reducing the impact of the unexpected faults.However,the existing static risk assessment methods provide limited information,which is not able to estimate the real-time risk change.Also,it cannot provide auxiliary decision-making for trains to operate at low risk.Therefore,this thesis proposes a dynamic risk assessment model based on operating safety boundary.The main work is as follows:(1)The safety boundary of train operation is defined and designed according to the characteristics of urban rail transit safety assurance.Based on the concept of Operational Design Domain(ODD)of intelligent vehicle to demarcate safe operation conditions,a train operation safety margin model was proposed,which was used to realize automatic derivation of safe operation scenarios.Time automaton UPPAAL is used to realize automaton modeling of safety boundary model.And safety scenes are generated automatically on this basis.The safety boundary model clarifies the key scene parameters that need to be collected during operation.(2)Key scene parameters are used to determine the scenario information in dynamic risk assessment that needs to be captured.Based on the safety boundary model,the structure of a dynamic Bayesian network(DBN)is derived from key parameters and their dependencies.Fuzzy set theory is introduced to solve the problem of uncertainty affecting reasoning.The triangular fuzzy number method is introduced in detail to solve the uncertainty of data probability.On this basis,a fuzzy dynamic Bayesian network(FDBN)based on monitoring scene data deviation is proposed.And the risk situation index(RSI)is defined to provide the necessary early warning of risks.The semi-quantitative evaluation of train operation risk is carried out with equipment status and environmental changes as input.(3)Based on the dynamic risk assessment model,a run-time safety monitoring framework(RSMF)is proposed.RSMF is used to monitor in real-time whether the functionality of the system exceeds the safety boundary.In this thesis,the main activities of RSMF during the design and runtime period as well as the design of functional modules are studied in detail.The engineering application of the run-time risk assessment model is realized by RSMF.(4)Several typical scenarios are selected from the actual data of the Yanfang Line of the Beijing metro for runtime risk assessment.In addition,the possible risk range,and the time and space margin of safe operating under different scenarios are deduced through the evaluation index of train dynamic risk behavior.The analysis results show that,when different external events occur,timely operational monitoring of safety risks can effectively perceive the trend of risk change.The validity of the evaluation model proposed in this thesis is verified.The model addresses the assessment problem of unclear risk changes during operation,illustrating its contribution to run-time risk assessment.There are 79 figures,27 tables,and 99 references in this thesis.