Abstract
Turnouts are the most expensive part to maintain on the railway track and therefore automated systems for detecting turnout defects are of great interest. Machine learning can improve predictive maintenance and is often used in automatic systems for precise prognosis. In this study, machine learning is used for identifying the condition of railway turnouts and potentially reducing costs by early automatic detection of defects. To train a machine learning algorithm, ordered, structured and categorized data (labelled data) are needed. A method is proposed to label the condition of turnouts in the Danish Railway based on a collection of repair records. This labelling of the turnouts is accomplished with unsupervised methods, namely a principal component analysis (PCA) followed by a cluster analysis. The labelling of the turnouts is investigated through comparisons of geometric measurements captured from the recording car. The difference in the physical properties illustrated by the geometric data indicates that the labelling is a good indicator of the relative condition of the turnout. When the data are labelled, supervised learning can be used to optimize the predictive power of machine learning algorithms (i.e. the algorithm learns from the labelled data) for classification of turnouts.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Kassa E, Nielsen JCO (2009) Dynamic train-turnout interaction in an extended frequency range using a detailed model of track dynamics. J Sound Vibr 320(4–5):893–914
Ma Y, Mashal AA, Markine VL (2018) Modelling and experimental validation of dynamic impact in 1:9 railway crossing panel. Tribol Int 118:208–226
Wang X, Chi Y, Li W, Zhou T (2012) Study on the dem simulation of the granular railway ballast bed tamping. Adv Mater Res 524–527(4):3256
Cornish A, Kassa E, Smith R (2012) Field experimentation and analysis at switches and crossings in UK. In: 9th International conference on contact mechanics and wear of rail/wheel systems, CM 2012, pp 649–651
Lee J, Choi H, Park D, Chung Y, Kim HY, Yoon S (2016) Fault detection and diagnosis of railway point machines by sound analysis. Sensors 16(4):549
Nielsen JCO, Palsson BA, Torstensson PT (2016) Switch panel design based on simulation of accumulated rail damage in a railway turnout. Wear 366:241–248
Pletz M, Daves W, Ossberger H (2012) A wheel set/crossing model regarding impact, sliding and deformation-explicit finite element approach. Wear 294:446–456
Grossoni I, Bezin Y, Neves SGM (2016) Optimization of support stiffness at a railway crossing panel. In: Civil-comp Proceedings, 110 p
Dhar S, Zhang Y, Xu R, Danielsen HK, Juul Jensen D (2017) Synchrotron x-ray measurement of residual strain within the nose of a worn manganese steel railway crossing. In: IOP conference series: materials science and engineering, 219(1), 012016 p
Schmitt L, Létourneaux F, De Keyzer I, Crompton P (2016) CAPACITY4RAIL: towards an affordable, innovative and high-capacity European Railway System for 2030/2050. Mater Infrastruct 2(5B):105–114, Wiley Blackwell
Bo Nielsen NFN (2008) Lars Forsting, Banenorm bn1-14-2: Projektering, tilstand og eftersyn af sporskifter og sporskæringer, 64 p
Klaus Bergman NF (2008) Lars Forsting, Banenorm bn2-15-2: Projektering, etablering, tilstand og vedligeholdelse af sporskifter og sporskæringer, 31 p
Jensen PJ, Fongemie R (2016) Sporteknik: danske jernbanespors opbygning, konstruktion og komponenter. Banedanmark
Stanislav Jovanović, Dragan Božović, Mirjana Tomičić-Torlaković (2014) Railway infrastructure condition-monitoring and analysis as a basis for maintenance management. Hrvatski savez gradjevinskih inženjera, Cm 2012:347–358
CEN (2003) Railway applications—track–track geometry quality Part 1: Characterisation of track geometry. European Committee for Standardization (CEN); 2003 13848-1. https://standards.globalspec.com/std/13275308/en-13848-1
Nissen A, Parida A, Kumar U (2010) Condition monitoring of railway switches and crossing by using data from track recording cars. Int J Comadem 13(3):36–41
Bergquist B, Soderholm P (2015) Data analysis for condition-based railway infrastructure maintenance. Qual Reliab Eng Int 31(5):773–781
Seyed Asadzadeh M, Galeazzi R, Hovad E, Fink Andersen J, Thyregod A, Camilla FS (2018) Rodrigues, Ballast degradation modeling for turnouts based on track recording car data. In: Proceedings of the european conference of the PHM society, , Vol 4 no 1
Kumar U, Ahmadi A, Verma AK, Prabhakar V (2016) Current trends in reliability, availability, maintainability and safety, current trends in reliability, availability, maintainability and safety. Springer International Publishing
Friedman J, Hastie T, Tibshirani R (2017) The elements of statistical learning, 2nd edn. Springer series in Statistics, New York
Hovad E, Thyregod C, Fink Andersen J, Christian BL, Peter Spooner M, Rodrigues AFS, Kjær Ersbøll B (2019) The effect of driving direction on spatially aligned track recording car measurements in turnouts. Int J Rail Transport 1–15, Taylor and Francis Ltd
Alpaydin E (2009) Introduction to machine learning. MIT press
Kassambara A (2017) Practical guide to principal component methods in R: unsupervised machine learning, vol 1. STHDA
MacQueen J (1967) Some methods for classification and analysis of multivariate observations. In: Le Cam LM, Neyman J (eds) Proceedings of the fifth berkeley symposium on mathematical statistics and probability, vol 1, pp 281–297. University of California Press, Berkeley, CA
Hopkins B, Skellam JG (1954) A new method for determining the type of distribution of plant individuals. Ann Bot 18(2):213–227
Kaufman L, Rousseeuw PJ (1990) Finding groups in data: an introduction to cluster analysis. Wiley, New York
Cleveland WS, Devlin SJ (1988) Locally weighted regression: an approach to regression analysis by local fitting. J Am Stat Assoc 83(403):596–610
R Core Team (2017) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix: Clustering
Appendix: Clustering
1.1 Clustering Frequency for the Different Turnouts
The general cluster frequency for the turnouts can be seen in Fig. 8. Fewer turnouts are typically assigned to Cluster 1, but some turnout has a high frequency of Cluster 1. Typically a high frequency in Cluster 1 belongs to turnouts which are exposed to a larger tonnage (used more) and thereby more frequent maintenance is needed for these turnouts. Specifically it can be seen that the turnouts with the numbers of 1623, 5803, 5804 and 5805 (shown with the gray boxes) are more frequently assigned to Cluster 1.
Cluster 2 small community which could be seen in Fig. 8, the small community was especially separated in the third dimension (Dim.3). This community consist of two older turnouts with the numbers of 857 and number 5289 (shown with the black boxes) having campaigns captured in the period of 15,145 to 17,110 days after the renewal day. These two turnouts only had a few repairs and tamping/grindings take place in the defined time windows.
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Vassos, G., Hovad, E., Duroska, P., Thyregod, C., da Silva Rodrigues, A.F., Clemmensen, L.H. (2021). Labelling the State of Railway Turnouts Based on Repair Records. In: Galeazzi, R., Kjartansson Danielsen, H., Kjær Ersbøll, B., Juul Jensen, D., Santos, I. (eds) Intelligent Quality Assessment of Railway Switches and Crossings. Springer Series in Reliability Engineering. Springer, Cham. https://doi.org/10.1007/978-3-030-62472-9_10
Download citation
DOI: https://doi.org/10.1007/978-3-030-62472-9_10
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-62471-2
Online ISBN: 978-3-030-62472-9
eBook Packages: EngineeringEngineering (R0)