Automatic Rail Flaw Localization and Recognition by Featureless Ultrasound Signal Analysis

  • Valentina SulimovaEmail author
  • Alexander Zhukov
  • Olga Krasotkina
  • Vadim Mottl
  • Anatoly Markov
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10934)


Ultrasound testing is a popular technique to find some hidden rail damages. In this paper we focus on the modern Russian railway flaw detectors, such as AVICON-14, which produce the results of ultrasound testing in the form of B-scan signals. We propose an approach simple enough to do fast automatic localization of B-scan signal segments, which could contain rail flaws. In order to recognize the selected segments as flaws of some kind or not flaws we apply SVM classifier jointly with DTW-based dissimilarity measure, specifically adapted by us to B-scan signals. To improve rail flaw localization and recognition quality we preprocess B-scan signals by applying some filter and making their convergence. Fast localization procedure jointly with CUDA implementation of B-scan segments comparison possesses the possibility to process big amounts of data. The experiments have shown that all rail flaws have been localized correctly and cross-validation ROC-score = 0.82 for the rail flaw recognition has been reached.


Russian railways Ultrasound testing B-scan Rail flaw localization Rail flaws recognition DTW SVM Featureless approach 



The results of the research project are published with the financial support of Tula State University within the framework of the scientific project № 2017-66PUBL.


  1. 1.
    Jimenez-Redondo, N., Bosso, N., Zeni, L., Minardo, A., Shubert, F., Heinicke, F., Simrothhubert, A.: Automated and cost effective maintenance for railway (ACEM-Rail). Proc. Soc. Behav. Sci. 48, 1058–1067 (2012)CrossRefGoogle Scholar
  2. 2.
    Jemec, V., Grum, J.: Automated non-destructive testing and measurement systems for rails. In: 10th European Conference on Non-Destructive Testing, Moscow (2010).
  3. 3.
    Markov, A.A., Shpagin, D.A.: Ultrasonic Rail Defectoscopy, 2nd edn, p. 284. Education – Culture, St-Petersburg (2013). (in Russian)Google Scholar
  4. 4.
    Alahakoon, S., Sun, Y.Q., Spiryagin, M., Cole, C.: Rail flaw detection technologies for safer, reliable transportation: a review. J. Dyn. Syst. Meas. Control 140(2), 020801 (2017). Paper No: DS-17-1110CrossRefGoogle Scholar
  5. 5.
    Markov, А.А., Antipov, А.G.: Magnetodynamic method for rail testing. V mire nauki №3(57), 66–71 (2012). (in Russian)Google Scholar
  6. 6.
    Markov, A.A., Kozyakov, A.B., Kuznetsova, E.A.: Decryption of defectogram of ultrasonic rail control. In: Practical Aid, St-Petersburg, p. 206 (2006). (in Russian)Google Scholar
  7. 7.
    Markov, A.A., Garaeva, V.S.: About acoustic contact at the area of rail bolt-on-joints. Put and putevoe hozyaistvo, № 12, pp. 15–17 (2008). (in Russian)Google Scholar
  8. 8.
    Markov, A.A., Kuznetsova, E.A., Antipov, A.G., Verevkin, A.Y.: A way of railway diagnostic. Patent for invention, № 2521095, 27 June 2014. (In Russian)Google Scholar
  9. 9.
    Lin, J., Luo, S., Li, Q., Zhang, H., Ren, S.: Real-time rail head surface defect detection: a geometrical approach. In: Proceedings of IEEE International Symposium on Industrial Electronics, pp. 769–774 (2009)Google Scholar
  10. 10.
    Gimy, J., Hyfa, N., Krishnan, R.: Rail flaw detection using image processing concepts - a review. IJERT 3(4) (2014). ISSN 2278-0181Google Scholar
  11. 11.
    Markov, A.A., Kozyakov, A.B., Kuznetsova, E.A., Shpagin, D.A.: Lost and new technologies of rail testing. Put and putevoe hozyaistvo, № 8, pp. 2–9 (2013). (in Russian)Google Scholar
  12. 12.
    Markov, A.A., Kuznetsova, E.A.: Rail defectoscopy. Signal’s formation and analysis. T. 2. Defectogram interpretation, p. 332. SPb UltraPrint (2014) (in Russian)Google Scholar
  13. 13.
    Cheprasov, D.N., Malenichev, A.A., Sulimova, V.V., Krasotkina, O.V., Mottl, V.V., Markov, A.A.: Railway ultrasonic defectogram missing data recovery on the basis of semi-global alignment. In: Machine Learning and Data Mining, T. 1, № 12, pp. 1731–1751 (2015). (in Russian)Google Scholar
  14. 14.
    Shilov, M.N.: Methodological and algorithmic foundation and software for registration and analysis of defectograms at ultrasonic rail inspection, p. 153. Ph.D. thesis, St-Petersburg (2007). (in Russian)Google Scholar
  15. 15.
    Fedorenko, D.B.: Problems of automatization of multichannel ultrasonic rail signals decryption. In: Proceedings of Radioelectronnye kompleksy mnogocelevogo naznacheniya, St-Petersburg, pp. 117–120 (2011). (in Russian)Google Scholar
  16. 16.
    Heckel, T., Thomas, H., Kreutzbruck, M., Ruhe, S.: High speed non-destructive rail testing with advanced ultrasound and eddy-current testing techniques. In: Indian National Seminar and Exhibition on Non-Destructive Evaluation, NDE (2009)Google Scholar
  17. 17.
    Jiao, S.X., Wong, S.B.: Development of an automated ultrasonic testing system. In: Proceedings SPIE, vol. 5852, pp. 480–486. Nanyang Technological University, Singapore. International Society of Optical Engineers, Singapore (2004)Google Scholar
  18. 18.
    Sun, M., Lin, X., Wu, Z., Liu, Y., Shen, Y., Feng, N.: Non-destructive photoacoustic detecting method for high-speed rail surface defects. In: Proceedings of IEEE International Instrumentation and Measurement Technology Conference, I2MTC, pp. 896–900 (2014)Google Scholar
  19. 19.
    Wang, D., Lu, L., Zhang, H.J.: Speech segmentation without speech recognition. In: International Conference on Multimedia and Expo, ICME 2003, vol. 1, pp. 405–408 (2003)Google Scholar
  20. 20.
    Kehagias, A., Nidelkou, E., Petridis, V.: A dynamic programming segmentation procedure for hydrological and environmental time series. Stoch. Environ. Res. Risk Assess. 20, 77–94 (2006)MathSciNetCrossRefGoogle Scholar
  21. 21.
    Fearnhead, P.: Exact Bayesian curve fitting and signal segmentation. IEEE Trans. Signal Process. 53(6), 2160–2166 (2005)MathSciNetCrossRefGoogle Scholar
  22. 22.
    Sasan, M., Sharif, B.S.: A nonlinear variational method for signal segmentation and reconstruction using level set algorithm. Signal Process. 86(11), 3496–3504 (2006)CrossRefGoogle Scholar
  23. 23.
    Duin, R.P.W., De Ridder, D., Tax, D.M.J.: Experiments with a featureless approach to pattern recognitionGoogle Scholar
  24. 24.
    Mottl, V., Dvoenko, S., Seredin, O., Kulikowski, C., Muchnik, I.: Featureless pattern recognition in an imaginary Hilbert space and its application to protein fold classification. In: Perner, P. (ed.) MLDM 2001. LNCS, vol. 2123, pp. 322–336. Springer, Heidelberg (2001). Scholar
  25. 25.
    Martens, R., Claesen, L.: On-line signature verification by dynamic time-warping. In: ICPR, pp. 38–42. IEEE (1996)Google Scholar
  26. 26.
    Wang, X., et al.: Experimental comparison of representation methods and distance measures for time series data. Data Min. Knowl. Discov. 1–35 (2010)Google Scholar
  27. 27.
    Verbovoy, V.: Metrics for sound signals comparison with subject to human hearing’s characteristics. In: Computer Graphic and Multimedia, № 3, pp. 2–10 (2005)Google Scholar
  28. 28.
    Malenichev, A., Sulimova, V., Krasotkina, O., Mottl, V., Markov, A.: An automatic matching procedure of ultrasonic railway defectograms. In: Perner, P. (ed.) MLDM 2014. LNCS, vol. 8556, pp. 315–327. Springer, Cham (2014). Scholar
  29. 29.
    Bellman, R., Kalaba, R.: Dynamic programming and modern control theory. In: Science, p. 118 (1969). Pattern Recognit. Lett. 18(11–13), 1159–1166 (1997)Google Scholar
  30. 30.
    Zhukov, A.S., Malenichev, A.A., Krasotkina, O,V., Sulimova, V.V.: Fast algorithm of railway ultrasonic defectograms matching. In: Proceedings of XVIII International Conference on DAMDID/RCDL 2016. Analytics and Data Management in Areas with Intensive Data Using, Yershovo, 11–14 October 2016. TORUS PRESS (2016)Google Scholar
  31. 31.
    Vapnik, V.N.: Statistical Learning Theory, p. 768. Wiley-Interscience, Hoboken (1998)zbMATHGoogle Scholar
  32. 32.
    Mottl, V., Lange, M., Sulimova, V., Yermakov, A.: Signature verification based on fusion of on-line and off-line kernels. In: 19th International Conference on Pattern Recognition, Florida, Tampa, December 2008Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  • Valentina Sulimova
    • 1
    Email author
  • Alexander Zhukov
    • 2
  • Olga Krasotkina
    • 3
  • Vadim Mottl
    • 1
    • 4
  • Anatoly Markov
    • 5
  1. 1.Tula State UniversityTulaRussia
  2. 2.Sberbank of RussiaMoscowRussia
  3. 3.Markov Processes InternationalSummitUSA
  4. 4.Computing Center of the Russian Academy of SciencesMoscowRussia
  5. 5.Radioavionica CorporationSaint-PetersburgRussia

Personalised recommendations