EURO 2025 Leeds
Abstract Submission

2842. Multi-Objective Railway Timetabling: A Comparison of Weighted Sum and Augmented Chebyshev

Invited abstract in session MA-10: Timetabling, stream Automated Timetabling.

Monday, 8:30-10:00
Room: Clarendon SR 1.06

Authors (first author is the speaker)

1. George Brooks
Computer Science, Swansea University

Abstract

British railways have experienced significant changes to passenger behaviours following the Covid-19 pandemic, with a decrease in commuter traffic and an increase in leisure travel. This has led to challenges in how an efficient timetable is devised and managed, often tackling multiple conflicting objectives. The most common approach for solving multiple-objective problems for an integer linear program is to use a weighted sum of the objectives, which is computationally fast and well suited to problems where the Pareto front is known to be convex. One key issue with weighted sum is that it is known not to be able to distinguish weakly dominated solutions, and therefore it is not possible to locate all solutions in a non-convex front. Other approaches such as augmented Chebyshev can discover these solutions while maintaining comparable performance in convex problems. In our work, we created a macroscopic representation of rail networks with competing objectives to test the performance of both weighted sum and augmented Chebyshev for multi-objective timetabling. We compare the number of solutions found, produced hypervolumes, and time complexity to analyse the benefit of moving from one approach to the other. We are currently working on incorporating demand in our framework and timetable symmetry, which are desirable features to both Train operating companies and passengers.

Keywords

Status: accepted

Back to the list of papers