Teilnahme an der "International Conference on the Practice and Theory of Automated Timetabling"

© Laudel

Vom 28. bis 31. August 2018 fand die 12.  "International Conference on the Practice and Theorie of Automated Timetabling" (PATAT 2018) in Wien statt. Auf dem Campus der TU Wien trafen sich Wissenschaftler, Verbands- und Industrievertreter aus Europa, Asien und den USA, um aktuelle Forschungsthemen im Timetabling-Bereich zu präsentieren und zu diskutieren. Unser Mitarbeiter Martin Scheffler hielt einen Vortrag im Stream "Timetabling in Public Transportation" zur Umlaufplanung im schienengebundenen Güterverkehr.

Real-life Decision Support for the Locomotive Assignment Problem with Heterogeneous Consists.

Arising from a practical problem in European rail freight transport
we present a decision support system for the locomotive assignment problem
(LAP) with heterogeneous consists. Based on a pre-planned train schedule
a set of locomotives (consists) has to be assigned to a number of trains at
minimal costs under consideration of several real-world requirements.

By predefining consists (as artificial locomotives) the model can be heuris-
tically solved as simple LAP (with heterogeneous locomotives). But defining
a set of suitable consists with a reasonable size becomes more different with
increasing number of locomotive types. By reducing the underlying space-time
network heuristic solutions including light-travelling can be achieved. The re-
duction can be done by focusing the space, time or a combination of both.
We discuss different strategies and running orders of predefining consists and
reducing the space-time network to generate a heuristic solution which is deal-
ing well with both issues. Subsequently, this can be used as initial solution for
the original problem on a complete space-time network. The resulting decision support system is tested on real world data as well as on self-generated instances.