Title: A clustering search approach for the train timetabling problem

Authors: Eggo Henrique Freire Pinheiro; Enrico Silva Miranda; Glaubos Clímaco; Alexandre C.M. de Oliveira

Addresses: Vale SA, São Luís, Brazil ' Departamento de Informática, Universidade Federal do Maranhão, São Luís, Brazil ' Departamento de Informática, Universidade Federal do Maranhão, São Luís, Brazil ' Departamento de Informática, Universidade Federal do Maranhão, São Luís, Brazil

Abstract: Freight railways are the major means of transportation of bulk material. Since for the last few years there has been a fast-growing demand and railway infrastructure capacity increasing is very expensive, the improvement of the train scheduling process is needed to ensure the quality of services. This work deals with the train timetabling problem (TTP) composed of mixed traffic railways - both cargo and passenger trains sharing the same resources with different priorities. We propose a new formulation for the TTP, which considers the parallel multi-track context and overtaking in a planning horizon for ongoing and just planned trains. Besides, a novel application of an evolutionary clustering search (ECS) is presented to solve large TTP instances. Finally, we have built a new set of instances derived from real-world scenarios. The findings encourage the future development of ECS-based expert systems that can provide information to decision-making teams of mining companies.

Keywords: genetic algorithm; evolutionary clustering search; ECS; metaheuristic; train timetabling problem; TTP.

DOI: 10.1504/IJLSM.2024.140397

International Journal of Logistics Systems and Management, 2024 Vol.48 No.4, pp.437 - 464

Received: 23 Jul 2021
Accepted: 24 Nov 2021
Published online: 07 Aug 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article