Using a general-purpose Mixed-Integer Linear Programming solver for the practical solution of real-time train rescheduling

•We propose an effective heuristic for real-time train rescheduling.•The heuristic uses a general-purpose Mixed-Integer Linear Programming solver.•Randomized variable fixing produces improved solutions for parallel runs.•The approach proved successful for real cases provided by an industrial company...

Full description

Saved in:
Bibliographic Details
Published in:European journal of operational research Vol. 263; no. 1; pp. 258 - 264
Main Authors: Fischetti, Matteo, Monaci, Michele
Format: Journal Article
Language:English
Published: Elsevier B.V 16.11.2017
Subjects:
Combinatorial optimization
Mixed-Integer Linear Programming
Railways optimization
Real-time optimization
Train rescheduling
Real-time optimization
Train rescheduling
Combinatorial optimization
Mixed-Integer Linear Programming
Railways optimization
ISSN:0377-2217, 1872-6860
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•We propose an effective heuristic for real-time train rescheduling.•The heuristic uses a general-purpose Mixed-Integer Linear Programming solver.•Randomized variable fixing produces improved solutions for parallel runs.•The approach proved successful for real cases provided by an industrial company. At a planning level, train scheduling consists of optimizing the routing and scheduling for a set of trains on a railway network. In real-time operations, however, the planned schedule constantly needs to be verified and possibly updated due to disruptions/delays that may require train rerouting or cancelation. In practice, an almost immediate reaction is required when unexpected events occur, meaning that trains must be rescheduled in a matter of seconds. This makes the time-consuming optimization tools successfully used in the planning phase completely inadequate, and ad-hoc (heuristic) algorithms have to be designed. In the present paper we develop a simple approach based on Mixed-Integer Linear Programming (MILP) techniques, which uses an ad-hoc heuristic preprocessing on the top of a general-purpose commercial solver applied to a standard event-based MILP formulation. A computational analysis on real cases shows that our approach can be successfully used for practical real-time train rescheduling, as it is able to deliver (almost) optimal solutions within the very tight time limits imposed by the real-time environment.
ISSN:0377-2217
1872-6860
DOI:10.1016/j.ejor.2017.04.057