Publication detail

MACHINE LEARNING BASED TRAIN TYPE IDENTIFICATION AT RAILROAD SWITCH USING VIBRATION

KRATOCHVÍLOVÁ, M.

Original Title

MACHINE LEARNING BASED TRAIN TYPE IDENTIFICATION AT RAILROAD SWITCH USING VIBRATION

Type

conference paper

Language

English

Original Abstract

This work concerns the use of machine learning to identify trains passing through S&C based on the acceleration signal measured in the track. This system can be use in the future, for example, to track changes in the stiffness of the bearing structure over time and thus predict the need for maintenance. Several methods of machine learning were compared based on their accuracy, time and computational demands for a given problem and the optimal method (Support Vector Machine) was implemented on real data. Because of the small amount of usable data, the bootstrapping method was used to generate training and test datasubsets.

Keywords

Machine learning, switches and crossings, acceleration, train, support vector machine

Authors

KRATOCHVÍLOVÁ, M.

Released

23. 1. 2020

Publisher

Econ Publishing s.r.o.

Location

Brno

ISBN

978-80-86433-73-8

Book

22. ODBORNÁ KONFERENCE DOKTORSKÉHO STUDIA

Edition

22

Pages from

211

Pages to

216

Pages count

937

URL

http://www.juniorstav.cz/wp-content/uploads/2020/02/Sbornik_Komplet_FINAL-uprava.pdf

BibTex

@inproceedings{BUT162430,
  author="Martina {Floriánová}",
  title="MACHINE LEARNING BASED TRAIN TYPE IDENTIFICATION AT RAILROAD SWITCH USING VIBRATION",
  booktitle="22. ODBORNÁ KONFERENCE DOKTORSKÉHO STUDIA",
  year="2020",
  series="22",
  pages="211--216",
  publisher="Econ Publishing s.r.o.",
  address="Brno",
  isbn="978-80-86433-73-8",
  url="http://www.juniorstav.cz/wp-content/uploads/2020/02/Sbornik_Komplet_FINAL-uprava.pdf"
}