Human Reliability Analysis Method Research For Rail Traffic Based On Improved CREAM

Rail traffic is an important part of transportation in our country and its safety is related to people’s lives and property. Reliability analysis is an important method to ensure the safety of rail operation. Human, as an important part of the rail traffic system, is very important to analyze its reliability. Although human reliability analysis(HRA) has achieved more mature theoretical research accomplishment and practical application, the HRA in rail traffic domain is still under development. Since rail traffic has the characteristics of complex traffic management, high traffic density and whole day operation, human reliability analysis method is introduced to the rail traffic domain. The main work of this paper is summarized as follows:Firstly, the existing human reliability analysis methods and quantitative reasoning methods such as fuzzy logic, evidential reasoning and Bayesian network were studied. To predict the error probability, the typical CREAM method was improved, in which, the CPCs evaluation rules were proposed, and the fuzzy CPC inputs were gived using fuzzy logic. After determining the confidence in IF-THEN rules using evidence reasoning, the calculation formula of human error probability was gived using Bayesian inference. The improved basic method can get precise values of human error probability instead of a probability interval.Secondly, the improved extended method was designed to calculate the human error probability more accurately and reasonably. For the improved extended method, the cognitive model in CREAM was revised after considering the effect of human internal factors on the cognitive function, and the CPC dependency rules was introduced. After the HIFs and its influence weights to cognitive function were determined, the accuracy for the calculation formula of human error probability was improved.Thirdly, the analysis software for improved CREAM basic method and extended method was designed by C# programming language, Matlab software and SQL database technology. This software can realize the automatic calculation and analysis for human error probability and it is convenient for practical application of the improved method.Finally, the train arrival and departure operations scenarios were taken as the examples, and the CRH driver and emergency response personnel were studied. Their error probabilities were calculated by the improved method of CREAM, and the validity for the improved method of CREAM was verified. On the basis of analyzing human reliability, the reasonable suggestions and measures were presented to improve the human reliability in rail traffic domain.