All Issue

2020 Vol.19, Issue 1 Preview Page

Research Article

Prediction of Track Quality Index (TQI) Using Vehicle Acceleration Data based on Machine Learning

차량가속도데이터를 이용한 머신러닝 기반의 궤도품질지수(TQI) 예측

Chanyong Choi1*
,
Hunki Kim2
,
Young Cheul Kim3
,
Sang-su Kim4
최 찬용1*
,
김 현기2
,
김 영철3
,
김 상수4
1Member, Principal Researcher, Northern Railroad Research Division, Korea Railroad Research Institute, 176 Cheoldobangmulgwan-ro, Uiwang-si, Gyeonggi-do 16105, Republic of Korea
2Member, Principal Researcher, Advanced Railroad Civil Engineering Division, Korea Railroad Research Institute, 176 Cheoldobangmulgwan-ro, Uiwang-si, Gyeonggi-do 16105, Republic of Korea
3Nonmember, Manager, Research Planning Department, Korail Research Institute Korea Railroad Corp., 240 Jungang-ro, Dong-gu, Daejeon city, Korea Po 34618, Republic of Korea
4Nonmember, Director, Research Institute AI Team, WISEiTECH co., Ltd.,C-Dong, 5th Floor, Pangyo Inovalley, 253 Pangyo-ro, Bundang-gu, Seongnam-si, Gyeonggi-do, Republic of Korea
* Corresponding Author
31 March 2020. pp. 45-53
PDF XML Citation
Abstract

철도분야에서도 계측자료를 바탕으로 머신러닝 기법을 이용하여 예측 분석하는 시도가 점차적으로 증가하고 있는 실정이다. 이 논문에서는 열차의 차상가속도 데이터를 기반으로 궤도의 품질을 결정하는 지표 중에 하나인 궤도품질지수를 머신러닝 기법을 활용하여 예측하였다. 머신러닝 기법으로 활용하고 있는 대표적인 3개의 모델로 궤도품질지수를 예측하여 가장 정확도가 높은 모델은 XGBoost으로 데이터셋에서 85% 이상의 예측정확도를 보였다. 또한 윤축과 대차의 z축의 진동가속도가 고저 궤도품질지수의 기여도가 높은 것으로 나타났으며, 이는 기존 연구결과와도 잘 일치하였다. 이러한 결과를 볼 때 단일 알고리즘인 서포터 벡터머신보다는 앙상블 알고리즘을 적용한 랜덤포레스트와 XGBoost이 정확도가 높은 것으로 판단된다. 따라서 머신러닝 기법에서 적용모델에 따라 정확도가 달라질 수 있기 때문에 차량진동가속도를 이용한 궤도품질지수를 예측하기 위해서는 앙상블 알고리즘을 가지는 모델을 적용하는 것이 적절할 것으로 판단된다.

There is an increasing tendency to try to make predictive analysis using measurement data based on machine learning techniques in the railway industries. In this paper, it was predicted that Track quality index (TQI) using vehicle acceleration data based on the machine learning method. The XGB (XGBoost) was the most accurate with 85% in the all data sets. Unlike the SVM model with a single algorithm, the RF and XGB model with a ensemble system were considered to be good at the prediction performance. In the case of the Surface TQI, it is shown that the acceleration of the z axis is highly related to the vertical direction and is in good agreement with the previous studies. Therefore, it is appropriate to apply the model with the ensemble algorithm to predict the track quality index using the vehicle vibration acceleration data because the accuracy may vary depending on the applied model in the machine learning methods.

Keywords
Machine learning
Track Quality Index
Vehicle acceleration
Measuring data
Track irregularity

References
1
Chen, T. and Guestrin, C. (2016), “XGBoost: A Scalable Tree Boosting System”, KDD’16, pp.785-794.
10.1145/2939672.2939785
2
Choi, I. Y., Lee, J. S., Um, J. H. and Kang, T. K. (2015), “Characteristics of Track Irregularities on Kyoungbu High-speed Line”, Journal of the Korean Society for Railway, Vol.18, No.1, pp.63-70. (in Korean)
10.7782/JKSR.2015.18.1.63
3
Cortes, C. and Vapnik, V. (1995), Support-vector networks,. Machine Learning, 20(3): pp.273.
10.1007/BF00994018
4
Gibert, X., Patel, V. and Chellappa, R. (2017), “Deep multitask learning for railway track inspection”, IEEE Transactions on Intelligent Transportation Systems, 18(1), pp.153-164.
10.1109/TITS.2016.2568758
5
Jeong, D. H. and Jeong, W. T. (2019), “Prediction of Rolling Noise Based on Machine Learning Technique Using Rail Surface Roughness Data”, Journal of the Korean Society for Railway, Vol.22 No.3, pp.209-217. (in Korean)
10.7782/JKSR.2019.22.3.209
6
Jung, H. and Park, M. S. (2018), A Study of Big data-based Machine Learning Techniques for Wheel and Bearing Fault Diagnosis, Korea Academy Industrial Cooperation Society, Vol.19, No.1, pp.75-84. (in Korean)
7
Kang, K. D. (2006), “A study of analysis method for the track geometry measuring data on high speed railway”, Journal of the Korean Society of Steel Construction, Vol.18, No.1, pp.47-51. (in Korean)
8
Lasisi, A. and Attoh-Okine, N. (2018), “Principal components analysis and track quality index: A Machine learning approach”, Transportation Research Part C 91, pp.230-248.
10.1016/j.trc.2018.04.001
9
Lee, J. S., Hwang, S. H., Choi, I. Y. and Choi, Y. T. (2018), “Comparative Study on Crack Detection of Sleepers with Region- and Semantic Segmentation-based Models of Deep Learning”, Journal of the Korean Society for Railway, Vol. 21, No.9, pp.924-932. (in Korean)
10.7782/JKSR.2018.21.9.924
10
Nakhaee, M. C. and Hiemstra, D. (2019), “The Recent Applications of Machine Learning in Rail Track Maintenance: A Survey”, International Conference on Reliability, Safety and Security of Railway Systems(RSSRail 2019), pp.91-105.
10.1007/978-3-030-18744-6_6
11
UIC 518 (2009), Testing and approval of railway vehicles from the point of view of their dynamic behaviour-Safety-Track fatigue-Ride quality.
12
Vapnik, V. (2000). Section 5.6. Support Vector Machines, The nature of statistical learning theory New York: Springer-Verlag New York. ISBN 978-1-4419-3160-3.
13
Woo, B. K., Kim, Y. H., Yun, U. S. and Kim, K. H. (2009), “Analysis of relationship between the irregularities of rail weld surface and track irregularities”, Proceeding of the Korean Society for Railway, Korea, pp.539-545. (in Korea)
Information
  • Publisher :Korean Geosythetics Society
  • Publisher(Ko) :한국지반신소재학회
  • Journal Title :Journal of the Korean Geosynthetics Society
  • Journal Title(Ko) :한국지반신소재학회 논문집
  • Volume : 19
  • No :1
  • Pages :45-53
  • Received Date : 2019-09-26
  • Revised Date : 2020-01-03
  • Accepted Date : 2020-02-26
  • DOI :https://doi.org/10.12814/jkgss.2020.19.1.045
Journal Informaiton Journal of the Korean Geosynthetics Society Journal of the Korean Geosynthetics Society
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close