10.14489/td.2026.09.pp.004-014

2026, 09 September

DOI: 10.14489/td.2026.09.pp.004-014

Bazulin A. E., Medvedev L. V.
DETECTION AND CLASSIFICATION OF INDICATIONS USING ECHO SIGNALS OBTAINED BY THE TIME-OF-FLIGHT DIFFRACTION METHOD WITH A CONVOLUTIONAL NEURAL NETWORK
(pp. 4-14)

Abstract. This paper proposes using a U-Net convolutional neural network to detect and classify indications in TOFD images. The focus is on the modeling and classification of reflectors such as internal and surface-breaking reflectors, pores, as well as the lateral wave signal and back-wall signal. The process of optimizing the neural network structure and the method for compiling and labeling the dataset are shown. The experiments involved training the model on TOFD images obtained from test blocks with model defects and from simulations in the CIVA software. The materials used to obtain the results have been compiled into a dataset that is publicly available.

Keywords: artificial neural network (ANN), convolutional neural network (CNN), time of flight diffraction (TOFD), ultrasonic testing (UT).

+ - About the Authors Click to collapse

A. E. Bazulin, L. V. Medvedev (ECHO+ Scientific and Production Center, LLC, Moscow, Russia) E-mail: Данный адрес e-mail защищен от спам-ботов, Вам необходимо включить Javascript для его просмотра. , Данный адрес e-mail защищен от спам-ботов, Вам необходимо включить Javascript для его просмотра.

+ - References Click to collapse

1. Skrynnikov, S. V. (2025). Concept of a modern system of automatic nondestructive testing of welded joints of main and field pipelines. Gazovaya promyshlennost, (1), 100–105. [in Russian language].
2. Bazulin, A. E., Kokolev, S. A., & Tikhonov, D. S. (2017). Attestation of the system and methodology of ultrasonic testing with full automation cycle. In 10th International Scientific and Technical Conference "Ensuring Safety of NPP with VVER" (p. 130). Podolsk. [in Russian language].
3. Ginzel, E. (2021). TOFD: Diffraction time of flight method of ultrasonic flaw detection: Basic principles and practical guide to application. DPK Press. [in Russian language].
4. Bazulin, A. E., Butov, A. V., Tikhonov, D. S., et al. (2020). Application of TOFD technology in developments of LLC "NPC "EHO+". Kontrol'. Diagnostika, (5), 28–37. [in Russian language].
5. Transneft. (2023). RD 25.160.10 KTN 0016 23: Main pipeline transport of oil and oil products. Nondestructive testing of welded joints during construction and repair of main pipelines. Transneft. [in Russian language].
6. Gazprom. (2023). STO Gazprom 15 2.3 005 2023: Welding and nondestructive testing of welded joints. Ultrasonic quality control of welded joints. Gazprom. [in Russian language].
7. Roca Barceló, F., Jaén del Hierro, P., Ribes Llario, F., & Real Herráiz, J. (2017). Development of an ultrasonic weld inspection system based on image processing and neural networks. Nondestructive Testing and Evaluation, 32(4), 325–340. https://doi.org/10.1080/10589759.2017.1376056
8. Shitole, C. N., Zahran, O., & Al Nuaimy, W. (2007). Combining fuzzy logic and neural networks in classification of weld defects using ultrasonic time of flight diffraction. Insight – Non Destructive Testing and Condition Monitoring, 49(2), 79–82.
9. Wang, Y., Cao, Y., Ye, B., et al. (2021). Research on detection method for welding seam defects in ultrasonic TOFD image based on Mask R CNN. Journal of Physics: Conference Series, 1995(1), 012032. https://doi.org/10.1088/1742-6596/1995/1/012032
10. cocodataset. (n.d.). cocodataset/cocodataset.github.io. GitHub. Retrieved February 6, 2026, from https://github.com/cocodataset/cocodataset.github.io.git
11. COCO Consortium. (n.d.). COCO Object in Context. Retrieved February 6, 2026, from https://cocodataset.org/#home
12. Huang, H., Hu, L., Li, B., et al. (2019). Recognition of defect in TOFD image based on Faster Region Convolutional Neural Networks. Welding, 41(7), 12–17. https://doi.org/10.11973/wsjc201907004
13. ImageNet. (n.d.). ImageNet. Retrieved February 6, 2026, from https://www.image-net.org/about.php
14. Bazulin, E. G., & Medvedev, L. V. (2025). Recognition of reflector type using a neural network based on TOFD echo signals. Defektoskopiya, (6). [in Russian language]. https://doi.org/10.31857/S0130308225060013
15. Ronneberger, O., Fischer, P., & Brox, T. (2015). U Net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer Assisted Intervention – MICCAI 2015 (pp. 234–241). Springer. https://doi.org/10.1007/978-3-319-24574-4_28
16. American Petroleum Institute. (1999). API 1104: Welding of Pipelines and Related Facilities. Retrieved February 6, 2026, from https://archive.org/details/gov.law.api.1104.1999/page/n35/mode/2up
17. American Society of Mechanical Engineers. (2010). ASME Boiler and Pressure Vessel Code, Section V: Nondestructive Examination. New York, NY: ASME. [in Russian language].
18. GOST ISO 10863 2022. (2022). Nondestructive testing of welded joints. Ultrasonic testing. Application of time of flight diffraction method (TOFD). Russian Institute of Standardization. [in Russian language].
19. Medved, L. (2026). TOFD dataset. Zenodo. https://doi.org/10.5281/zenodo.18471254

+ - Purchase digital version of a single article Click to collapse

This article is available in electronic format (PDF).

The cost of a single article is 750 rubles. (including VAT 20%). After you place an order within a few days, you will receive following documents to your specified e-mail: account on payment and receipt to pay in the bank.

After depositing your payment on our bank account we send you file of the article by e-mail.

To order articles please copy the article doi:

10.14489/td.2026.09.pp.004-014

and fill out the form