A data-driven time supplements allocation model for train operations on high-speed railways

This paper presents a time supplements allocation (TSA) method that incorporates historical train operation data to optimize buffer-time distribution in the sections and stations of a published timetable. First, delay recovery behavior is investigated and key influential factors are identified using real-world train movement records from the Wuhan-Guangzhou High-speed Railway (WH-GZ HSR) in China. Then, a ridge regression model is proposed that explains delay recovery time (RT) regarding buffer times at station (BTA), buffer times in section (BTE), and the severity of the primary delay (PD). Next, a TSA model is presented that takes the quantitative effects of identified factors as input to optimize time supplements locally. The presented model is applied to a case study comparing the existing and optimized timetables of 24 trains operating during peak morning hours. Results indicate an average 12.9% improvement in delay recovery measures of these trains.