Risk Analysis of Train Control System Based on Fuzzy Mathematics and Bayesian Network

Abstract

This paper takes the ‘train overspeed unprotected’ event of Chinese train control system level 2 (CTCS-2) as the studied risk event and analyzes the weak links leading to the risk event. Fault tree, expert scoring, fuzzy mathematics, and Bayesian network (BN) are applied in combination to conduct this study. Firstly, by analyzing the structure and function of CTCS-2 system, all the possible risk sources are identified, and a fault tree is established by taking the ‘train overspeed unprotected’ as the top event. Secondly, based on expert scoring method, a calculation model is established by using fuzzy mathematics to calculate the frequencies of risk sources. Thirdly, the fault tree is transformed into a BN, and the frequency of risk sources is used as the input of BN leaf nodes. The frequency of the risk event and the posterior probability of each risk source are calculated to identify the main weak links that lead to the risk event. The three main risk sources are track circuit, balise transmission module (BTM), and train control center (TCC), which provides the basis for taking further decision support for risk reduction control measures.