TU Braunschweig

Symbolfoto

Diaz Fuentes, D. E.; Grasso Toro, F.; Hudik, M.; Schnieder, E.:
Optimisation of Railway Maintenance Scheduling by Parallel Computing.
ECT 2014 - The Ninth International Conference on Engineering Computational Technology, Neapel, Italien, September 2014.

Kurzfassung:

The need for modern approaches to automate planning and scheduling for maintenance operations in the railway sector has become very relevant in the recent times as a result of the imminent European Train Control System (ETCS) migration strategies. In the framework of the AUTOMAIN project (Augmented Usage of Track by Optimisation of Maintenance, Allocation and Inspection of railway Networks) research has been conducted into optimisation algorithms for automated maintenance planning and scheduling resulted in a prototype with a three separated modules architecture. The interconnectivity and performance between modules present remarkable opportunities for parallel optimisation. The identification of the relevant characteristics of several parallel computing approaches related to the performance of the scheduling algorithms resulted in the selection of the Message Passing Interface (MPI) as the best optimising approach. This paper shows the evaluation of the MPI grid-based optimisation by means of number of parallelisable calculation calls and parallel computational time estimation. The achievable reduction in the total calculation time using the developed tool for planning and scheduling is presented here with respect to speed-up and efficiency. A numerical example for the validation process the automated railway maintenance planning and scheduling tool is presented.


http://www.iva.ing.tu-bs.de?iT=27_815&id_select=1121