Use of a genetic algorithm to improve the rail profile on Stockholm underground

Ingemar Persson, Rickard Nilsson, Ulf Bik, Magnus Lundgren, Simon Iwnicki

Research output: Contribution to journalArticle

24 Citations (Scopus)

Abstract

In this paper, a genetic algorithm optimisation method has been used to develop an improved rail profile for Stockholm underground. An inverted penalty index based on a number of key performance parameters was generated as a fitness function and vehicle dynamics simulations were carried out with the multibody simulation package Gensys. The effectiveness of each profile produced by the genetic algorithm was assessed using the roulette wheel method. The method has been applied to the rail profile on the Stockholm underground, where problems with rolling contact fatigue on wheels and rails are currently managed by grinding. From a starting point of the original BV50 and the UIC60 rail profiles, an optimised rail profile with some shoulder relief has been produced. The optimised profile seems similar to measured rail profiles on the Stockholm underground network and although initial grinding is required, maintenance of the profile will probably not require further grinding.

Original languageEnglish
Pages (from-to)89-104
Number of pages16
JournalVehicle System Dynamics
Volume48
Issue numberSUPPL. 1
DOIs
Publication statusPublished - 2010
Externally publishedYes

Fingerprint

Rails
Genetic algorithms
Wheels
Fatigue of materials
Computer simulation

Cite this

Persson, Ingemar ; Nilsson, Rickard ; Bik, Ulf ; Lundgren, Magnus ; Iwnicki, Simon. / Use of a genetic algorithm to improve the rail profile on Stockholm underground. In: Vehicle System Dynamics. 2010 ; Vol. 48, No. SUPPL. 1. pp. 89-104.
@article{27d41a17e2b34b9faba2a178766f691f,
title = "Use of a genetic algorithm to improve the rail profile on Stockholm underground",
abstract = "In this paper, a genetic algorithm optimisation method has been used to develop an improved rail profile for Stockholm underground. An inverted penalty index based on a number of key performance parameters was generated as a fitness function and vehicle dynamics simulations were carried out with the multibody simulation package Gensys. The effectiveness of each profile produced by the genetic algorithm was assessed using the roulette wheel method. The method has been applied to the rail profile on the Stockholm underground, where problems with rolling contact fatigue on wheels and rails are currently managed by grinding. From a starting point of the original BV50 and the UIC60 rail profiles, an optimised rail profile with some shoulder relief has been produced. The optimised profile seems similar to measured rail profiles on the Stockholm underground network and although initial grinding is required, maintenance of the profile will probably not require further grinding.",
keywords = "flange climb ratio, genetic algorithm, noise, optimisation, penalty function, rail profiles, RCF, Stockholm underground, time-domain simulations, track-shift forces, unevenly distributed wear, wear",
author = "Ingemar Persson and Rickard Nilsson and Ulf Bik and Magnus Lundgren and Simon Iwnicki",
year = "2010",
doi = "10.1080/00423111003668245",
language = "English",
volume = "48",
pages = "89--104",
journal = "Vehicle System Dynamics",
issn = "0042-3114",
publisher = "Taylor and Francis Ltd.",
number = "SUPPL. 1",

}

Use of a genetic algorithm to improve the rail profile on Stockholm underground. / Persson, Ingemar; Nilsson, Rickard; Bik, Ulf; Lundgren, Magnus; Iwnicki, Simon.

In: Vehicle System Dynamics, Vol. 48, No. SUPPL. 1, 2010, p. 89-104.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Use of a genetic algorithm to improve the rail profile on Stockholm underground

AU - Persson, Ingemar

AU - Nilsson, Rickard

AU - Bik, Ulf

AU - Lundgren, Magnus

AU - Iwnicki, Simon

PY - 2010

Y1 - 2010

N2 - In this paper, a genetic algorithm optimisation method has been used to develop an improved rail profile for Stockholm underground. An inverted penalty index based on a number of key performance parameters was generated as a fitness function and vehicle dynamics simulations were carried out with the multibody simulation package Gensys. The effectiveness of each profile produced by the genetic algorithm was assessed using the roulette wheel method. The method has been applied to the rail profile on the Stockholm underground, where problems with rolling contact fatigue on wheels and rails are currently managed by grinding. From a starting point of the original BV50 and the UIC60 rail profiles, an optimised rail profile with some shoulder relief has been produced. The optimised profile seems similar to measured rail profiles on the Stockholm underground network and although initial grinding is required, maintenance of the profile will probably not require further grinding.

AB - In this paper, a genetic algorithm optimisation method has been used to develop an improved rail profile for Stockholm underground. An inverted penalty index based on a number of key performance parameters was generated as a fitness function and vehicle dynamics simulations were carried out with the multibody simulation package Gensys. The effectiveness of each profile produced by the genetic algorithm was assessed using the roulette wheel method. The method has been applied to the rail profile on the Stockholm underground, where problems with rolling contact fatigue on wheels and rails are currently managed by grinding. From a starting point of the original BV50 and the UIC60 rail profiles, an optimised rail profile with some shoulder relief has been produced. The optimised profile seems similar to measured rail profiles on the Stockholm underground network and although initial grinding is required, maintenance of the profile will probably not require further grinding.

KW - flange climb ratio

KW - genetic algorithm

KW - noise

KW - optimisation

KW - penalty function

KW - rail profiles

KW - RCF

KW - Stockholm underground

KW - time-domain simulations

KW - track-shift forces

KW - unevenly distributed wear

KW - wear

UR - http://www.scopus.com/inward/record.url?scp=78649722980&partnerID=8YFLogxK

U2 - 10.1080/00423111003668245

DO - 10.1080/00423111003668245

M3 - Article

VL - 48

SP - 89

EP - 104

JO - Vehicle System Dynamics

JF - Vehicle System Dynamics

SN - 0042-3114

IS - SUPPL. 1

ER -