Correlation Analysis and Application of Railway Accident Risk Factors Fusing Experience and Data

Previous researchers focused on the analysis of the risk factors themselves and their role in the system when analyzing the risk factors of railway accidents, ignoring the analysis of the correlation between risk factors. The correlation analysis of railway accidents helps to understand the evolution process of risk factors at the micro level. This paper uses the DEMATEL method to fuse experience and data to quantify the correlation degree of railway accident risk factors, then build a risk factor correlation network. According to the edge weights of nodes in the correlation network, the most relevant adjacent nodes can be deduced, therefore potential faults can be inferred from known faults; important risk associations can be filtered in the associated network through thresholds, and risk propagation paths can be extracted from the simplified network. risk transmission links. Apply the method proposed in this paper to the railway equipment accident database of FRA (Federal Railroad Administration), establish a risk correlation network, give 4 potential risk reasoning cases, and extract 7 risk propagation paths. Analyzing the risk propagation paths, the results show that human factors such as driver's operation error and signal instruction transmission error are the most likely to be propagated and evolved into accidents as fundamental risk factors.