Mining railway traffic control logs

Felix Mannhardt (Corresponding author), Andreas D. Landmark

Research output: Contribution to journalConference articlepeer-review

8 Citations (Scopus)

Abstract

Railway traffic is a set of interrelated processes that are centrally controlled. Despite optimized train schedules, train dispatchers still take ad-hoc decisions on the scheduling of trains in the context of unplanned events. Train orders are swapped, train crossings on single-tracks are moved, or trains are cancelled to minimize the disruption in the schedule. The actual scheduling of trains, as decided by dispatchers and observed through the movement of trains across stations, is then registered in railway traffic control logs. Using this data that contains information on the tacit knowledge of dispatchers can help to evaluate strategies for dealing with disruptions, which have not been subject to upfront planning. This paper proposes to use process mining methods, which are commonly applied in the context of business processes, to expose the hidden process of how the train traffic was actually dispatched. Different variants of dispatching are juxtaposed with the total delay in the railway system to visually explore the dispatching strategies taken. The technique has been implemented as a prototype and validated on a large dataset of real-life traffic in the Norwegian railway system.

Original languageEnglish
Pages (from-to)227-234
Number of pages8
JournalTransportation Research Procedia
Volume37
DOIs
Publication statusPublished - 2019
Externally publishedYes
Event21st EURO Working Group on Transportation Meeting, EWGT 2018 - Technische Universitat Braunschweig, Braunschweig, Germany
Duration: 17 Sept 201819 Sept 2018

Funding

This work was in part supported by Bane NOR.

Keywords

  • Process mining
  • Railway
  • Traffic control logs
  • Traffic disruptions

Fingerprint

Dive into the research topics of 'Mining railway traffic control logs'. Together they form a unique fingerprint.

Cite this