Abstract
Efficient route optimization plays a vital role in ensuring both safety and punctuality in railway operations. This is particularly crucial in heterogeneous multi-gauge railway networks where varying train speeds, stopping patterns, and infrastructure compatibility constraints increase coordination complexity. In single-track systems, these challenges are intensified as all trains share the same track, necessitating frequent track switching. Stochastic disruption events, including blocked tracks, train failures, engine failures, and speed slowdowns, introduce additional unpredictability in operations and deviate the timetable. However, existing studies predominantly focus on high-level timetabling, omitting operational details such as track switching coordination, which leaves decision-making to human operators and increases safety risks in railway operations.
This study proposes a framework based on temporal planning for dynamic route optimization and disruption management in heterogeneous railway systems. The framework formulates railway operations as a temporal planning problem using PDDL 2.1, explicitly modeling gauge compatibility constraints and diverse disruption scenarios. It generates conflict-free timestamped operational plans specifying both optimized schedules and executable action sequences. To evaluate the proposed framework, we developed a benchmark problem set with 200 instances using up to 1,000 track points and 120 trains. Two state-of-the-art temporal planners and a plan validator were employed to assess the framework. Experimental results demonstrate that the framework effectively generates temporal operational plans for heterogeneous railway systems, handles multi-gauge constraints and disruptions, and reduces dependence on manual decision-making.
Blogger's Review: The proposed dynamic route optimization framework provides an innovative solution to the complex scheduling issues in railway systems by incorporating disruption factors, significantly enhancing system safety and efficiency. Future research could further explore the integration of machine learning with this framework to improve its adaptability and intelligence.