Quantitative Human Error Probability Prediction Model Designment And Risk Evaluation For Railway

In safety-critical system,human is the important factor that affects system safety.And human errors have become the main causes of accidents.It has an extremely important realistic meaning on researching human error mechanism,building human error probability prediction model and conducting risk evaluation for different human errors.Driven by the requirements from rail transport industry human reliability analysis(HRA),this dissertation is devoted to the further study of human error in railway by building model-based human error probability prediction method and putting forward quantitative analysis method.At the same time,human error risk evaluation method is proposed to provide reference for accurate risk control.The main work of this dissertation is as follows:1.The implementation means of model-based HRA method for railway is proposed and the quantitative analysis method is designed.In the aspect of qualitative analysis,human error probability prediction model is constructed under the guidance of framework of model-based HRA method structure.In crew cognitive behavior layer,THERP-HRA event tree based on hierarchy task analysis is introduced to model crew cognitive process.In crew failure mode layer,the CFM(Crew Failure Mode)identification method is proposed by introducing human error classification theory.And in performance shaping factor layer,hierarchy Bayesian networks based on SHELL model and Interpretative Structural Model is constructed.In the aspect of quantitative analysis,prior probabilities distribution of the root nodes of Bayesian network is achieved using D-S evidence theory in order to overcome the limitation of data and avoid experts’ evidence conflict.An automatic CPT(Conditional Probability Table)assignment method for intermediate nodes of Bayesian network in HRA based on fuzzy inference is designed to solve the problem of CPT assignment amount explosion.Quantitative connection between performance shaping factors and crew failure modes is realized by proposed method based on SLIM method.2.Risk evaluation indexes are summarized through analyzing the characteristics of human errors in railway.And a risk evaluation model based on VIKOR method improved by cloud model is designed.Considering the fuzziness and randomness in the process of experts’ evaluation,the natural language is transferred into cloud model to consist cloud decision matrix.Then the human errors can be ranked using VIKOR method based on cloud model.The key human errors can be identified and priority of risk control will be determined.3.Human error probability prediction hierarchy model is applied to analyze emergency response of dispatcher in high speed railway track red light appearance scenario.Failure rate of each cognitive behavior phase and the final HEP is calculated using the model quantitative analysis method.And validation of the method is carried out by sensitive analysis.In the end,the risk of human errors of dispatcher are ranked through risk evaluation method.The results can provide reference for railway departments in risk control decision making.