Optimizing Rail Shunting Operations through Multi-Stage AI Heuristics

Research output: Contribution to conferencesPresentation slidesContributed

Abstract

The efficiency of shunting operations in flat yards is crucial to the performance of railway networks, especially in single-wagon systems. Inefficiencies in these operations can lead to significant delays and increased operational costs. This impacts the overall reliability of rail systems. Despite the critical nature of this challenge, existing methods often rely on manual interventions.
We propose a novel algorithm, HEROS, tailored for optimizing shunting operations in flat yards. HEROS integrates elements of reinforcement learning and evolutionary algorithms, offering a sophisticated combination of AI concepts. Our algorithm is highly configurable, allowing it to be adapted to the specific needs of different flat rail yards.
Results show that HEROS consistently improves the efficiency of shunting operations, with the objective value converging and the standard deviation decreasing as the time budget increases. This indicates increased robustness and reliability over time. The algorithm offers substantial benefits in terms of cost savings, reliability, and scalability, making it a valuable tool for rail operators facing operational complexity.

Details

Original languageEnglish
Publication statusPublished - 20 Oct 2024
Peer-reviewedNo

Conference

Title2024 INFORMS Annual Meeting
SubtitleSmarter Decisions for a Better World
Abbreviated titleINFORMS 2024
Duration20 - 23 October 2024
Website
LocationSeattle Convention Center & The Hyatt Regency Seattle
CitySeattle
CountryUnited States of America