Human Reliability Analysis Of Locomotive Driving Process Based On Empirical Data

The railway system is an important means of transportation.With frequent occurrence of accidents and unsafe factors,the state economy has suffered huge losses.Despite gradual improvement of railway system technology has enhanced reliability of equipment,high standards of integration and automation raise the difficulties of human-machine interaction.However,the reliability of manual operation has not enhanced,which reflects mostly on mistakes of employees and organizational management defects.There are many factors causing incidents,such as human factors,organizations,management,environment and the combination of these factors mentioned above.It is crucial for accident investigation report as ex post management mechanism to acknowledge and analyze the causes and process of each incident.Unfortunately,accident investigation report is just a kind of qualitative analysis,not reflecting on the general law and mechanism of accident formation.If the structured accident data can be extracted from accident report,by means of theories and methods of computer and systems engineering,analyzing the main causes of the accident from empirical data,to reveal the objective law between human reliability and accident,is worthwhile for prior prevention and control of operations.This paper sets human reliability analysis of locomotive driving process as examples,comprehensively using the theory of literature analysis,empirical analysis,correlation analysis and complex network,directed by accident cause theory and human reliability theory,to put forward a quantitative model of the probability of human error in railway.Firstly,combined with the railway regulations and accident reports to slightly modify the standard HFACS(Human Factor Analysis Classification System)framework,establishment of a HFACS-railroad industry(HFACS-RR)framework for the railway system makes irregular text report causes sense.Besides,based on HFACS-RR,611accident/incident reports were composed to construct the railway accident human factor database.By means of Chi-squared Test and Lambda Correlation Analysis,we quantitatively calculate adjacent layer factors’ relationships among "unsafe acts of operators","precondition for unsafe behavior","unsafe supervision" and "organizational influence" to figure out the key factor of the accident and the key causal links,revealing complex dependence between human and organization factors.Thirdly,taking the difficulty in quantifying the correlations between behavior formation factors and other people and organizational factors in traditional SLIM(Success Likelihood Index Method)into account,we use the Complex Network theory to abstract the network of railway accident and organizational factors.Also,we describe the degree of importance from point in-degree,weight strength,node efficiency and propose solution flow.Finally,using SLIM combined with specific cases calculates human error probability during driving and constructs fault tree for human error tasks.Then this method is verified by stimulation by means of Monte Carlo method with uncertain parameters.The human reliability model presented in this paper has the ability of quickly identifying the human and organizational factors of accident report,quantifying the reliability of human,providing theoretical support for railway accident safety management and providing guidance to train railway personnel.