Research On The Causation Of Railway Accident Based On Complex Network Theory

The railway system in China has developed rapidly in recent years. In particular, the mileage of high-speed railway is expected to reach160,000kilometers in2020, an objective written in a governmental project document for the national high-speed railway development in the "12th five-year plan" period issued by the Ministry of Science and Technology in China. In order to ensure railway safety, risk prevention and early warning are playing a particular important role for the fast progression of railway system.Here, this paper establishes a railway accident causation model with the framework of the complex network theory. By analyzing the model with cascading safe method which improved based on the cascading failure method, we evaluate the importance of potential factors for railway accidents, and recognize several key factors as powerful diagnostic and prognostic markers for railway accidents. The main content of this paper are as follows:(1) Formulate a railway accident causation model. Complex network theory is an effective method to evaluate the importance of nodes in a system and to determine the propagation paths among them. In this paper, we respectively regard railway causation factors and relationships among factors as nodes and edges in the network to simulate the railway accident. Then, some key factors which led to railway accidents can be found out though analyzing statistical characteristics of the network, such as degrees of nodes, degree distribution, shortest path, and the betweenness. Take7.23Yong-Wen railway accident as an example, analyzing the model and finding that the problem of siginal is the key factor of this accident.(2) Propose the cascading safe method. Here, a cascading safe method, an improved version for a canonical analytic approach, is proposed based on the railway accident causation model combined with the actual requirement of railway accidents. The initial load, capacity, and dynamic balance of the node are redefined by giving each node values for extra protection and safety threshold to describe the dynamic distribution of extra protection. Efficiency of the network is proposed to represent the results of cascading safe process and its effects on railway accident.(3) Analyze the model by using the proposed cascading safe method. This paper describes how to construct railway accident causation models and how to analyze the statistical characteristics based on the7.23Yong-Wen railway accidents and the data of railway accidents collected by FRA. By employing the cascading safe method proposed in this paper, we analyze the cascading safe process of the Hub node and the efficiency of each node in these models. Moreover, we single out the key factors for railway accidents and then evaluate the influences of different combination of manpower and material resources on railway accidents.