Estimating the Total Delay Times Owing to Disruptions in High-Speed Rail Systems

Accurately estimating the influence of disruptions is one of the key factors to improve real-time train dispatching and operation management abilities in High-Speed Railway (HSR). In this paper, delayed train groups were extracted from the historical train operation data of Wuhan-Guangzhou HSR in China, to analyze the total delay times (TDT) resulting from disruptions. In order to establish accurate forecasting models, we first classified disruptions into three clusters using a K-Means algorithm, considering the timetable and disruption characteristics. Then, parametric and non-parametric estimation approaches were applied to fit the distributions of TDT of each category. Next, a Kolmogorov-Smirnov testing method was employed to test the goodness-of-fit of the probability density models and select the best model for each cluster. Finally, the validating results show that the proposed models accurately revealed the characteristics of disruptions, and that these models can be used in real-time dispatching to predict the TDT.