Improved distributed model predictive control for rescheduling of railway traffic by manipulation of the cost functions

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review

Abstract

In this paper we introduce two distributed model predictive control (DMPC) approaches that significantly improve the quality of the solutions found compared to the DMPC approaches that were introduced by Kersbergen et al. [2014b] for the rescheduling of railway traffic, while the computation time only increased by a small fraction. In DMPC the global rescheduling problem is split up into several local problems that are solved by local model predictive controllers that communicate with each other to achieve a solution for the global rescheduling problem. We improve the solution found by the DMPC approaches by adjusting the weights in the local problems such that the delay propagation through the network is reduced. We compare the performance in terms of computation time and delay reduction of the different DMPC approaches with the global model predictive control approach for different lengths of the prediction horizon.
Original languageEnglish
Title of host publicationProceedings of the 6th International Seminar on Railway Operations Modelling and Analysis (RailTokyo2015)
Subtitle of host publicationtechnical Report 15-016
Number of pages12
Publication statusPublished - 2015
EventRailTokyo 2015, 6th International Conference on Railway Operations Modelling and Analysis: Planning, Simulation and Optimisation Approaches - Tokyo, Japan
Duration: 23 Mar 201526 Mar 2015

Publication series

NameTechnical reports DCSC _ Delft University of Technology
PublisherDCSC
Number15-016

Conference

ConferenceRailTokyo 2015, 6th International Conference on Railway Operations Modelling and Analysis
Abbreviated titleRailTokyo2015
CountryJapan
CityTokyo
Period23/03/1526/03/15

Fingerprint Dive into the research topics of 'Improved distributed model predictive control for rescheduling of railway traffic by manipulation of the cost functions'. Together they form a unique fingerprint.

Cite this