Timetable merging for the Periodic Event Scheduling Problem

The authors propose a new mixed integer programming based heuristic for computing new benchmark primal solutions for instances of the PESPlib. The PESPlib is a collection of instances for the Periodic Event Scheduling Problem (PESP), comprising periodic timetabling problems inspired by real-world railway timetabling settings, and attracting several international research teams during the last years. The authors describe two strategies to merge a set of good periodic timetables. These make use of the instance structure and minimum weight cycle bases, finally leading to restricted mixed integer programming formulations with tighter variable bounds. Implementing this timetable merging approach in a concurrent solver, the authors improve the objective values of the best known solutions for the smallest and largest PESPlib instances by 1.7 and 4.3 percent, respectively.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01850152
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jun 27 2022 5:16PM