Cui, Junfeng (Beijing National Railway Research and Design Institute of Signal), Yuan, Yang (Southwest Jiaotong University), SHI, Jing (Beijing National Railway Research & Design Institute of Signal &), Liu, Ling (Beijing National Railway Research and Design Institute of Signal), Li, Jie (Chengdu University of Information Technology), Huang, Ping (Southwest Jiaotong University)

A Hybrid Data-Driven Model for Estimating the Effect of Train Traffic Control Strategies in Real-Time Train Dispatching

Scheduled for presentation during the Invited Session "S16c-Control, Communication and Emerging Technologies in Smart Rail Systems" (TH-LA-T16), Thursday, November 20, 2025, 16:00−16:20, Southport 1

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

Train overtaking is a common train traffic control strategy for delayed trains to reduce the influence of train delays. The present study aims to provide train traffic controllers with the effects of the train overtaking strategy, by predicting its influences on train operations. We first identify factors affecting the effects of train overtaking. Based on the operation data of high-speed railways, a comprehensive analysis of train overtaking and its impacts on train operations was conducted. Then, the highly related factors were used to construct a predictive model for the departure delays of trains after the overtaking strategy was adopted. Based on the data characteristics among different features and variables, suitable neural network modules are selected to process different variables. Therefore, a hybrid neural network model for predicting the influences of overtaking was established. To validate the effectiveness of the predictive model, it is compared with commonly used predictive models in existing studies. The results indicate that the predictive performance and accuracy of the model in this paper are superior to the other three standard prediction models, incapable of providing an accurate estimation of the effects of train overtaking strategies on train traffic controllers.