A method for modeling and analysis of directed weighted accident causation network (DWACN)

作者：Zhou Jin; Xu Weixiang^*; Guo Xin; Ding Jing

来源：Physica A: Statistical Mechanics and Its Applications , 2015, 437: 263-277.

DOI：10.1016/j.physa.2015.05.112

• Accident analysis
• Directed weighted accident causation network (DWACN)
• Statistical properties of directed weighted complex network
• Event chains

Using complex network theory to analyze accidents is effective to understand the causes of accidents in complex systems. In this paper, a novel method is proposed to establish directed weighted accident causation network (DWACN) for the Rail Accident Investigation Branch (RAIB) in the UK, which is based on complex network and using event chains of accidents. DWACN is composed of 109 nodes which denote causal factors and 260 directed weighted edges which represent complex interrelationships among factors. The statistical properties of directed weighted complex network are applied to reveal the critical factors, the key event chains and the important classes in DWACN. Analysis results demonstrate that DWACN has characteristics of small-world networks with short average path length and high weighted clustering coefficient, and display the properties of scale-free networks captured by that the cumulative degree distribution follows an exponential function. This modeling and analysis method can assist us to discover the latent rules of accidents and feature of faults propagation to reduce accidents. This paper is further development on the research of accident analysis methods using complex network.

• 出版日期2015-11-1
• 单位北京交通大学