Automatic Welding, 2026, №02

2026 №02 (05)

DOI of Article 10.37434/as2026.02.06

2026 №02 (07)

---

"Avtomatychne Zvaryuvannya" (Automatic Welding), #2, 2026, pp. 47-56

Investigation of the effectiveness of neural networks in welding of critical structures for reliability enhancement

A.S. Novodranov, V.O. Koliada

E.O. Paton Electric Welding Institute of the NAS of Ukraine 11 Kazymyr Malevych Str., 03150, Kyiv, Ukraine. E-mail: artur19940731@gmail.com

The transition from foundry production of load-bearing parts of freight car bogies to alternative technologies, in particular electric arc welding, is a promising direction of development. This study considers the side frame of a freight car bogie. The connection of frame elements made of high-quality sheet metal by electric arc welding ensures compliance with the basic dimensions within ±1 mm. In addition, it allows reducing wear on wheelsets, as well as reducing the weight of unsprung masses. However, the effectiveness of manual welding depends largely on the qualifications of the welder. Defects may form as a result of welding technology violations. Such limitations create the prerequisites for the use of automated welding equipment based on robotic complexes. Robotisation of welding production ensures high productivity, as well as compliance with all technological requirements. However, the use of welding robots alone cannot ensure defect-free production, which prompts the integration of automated non-destructive testing (NDT) systems. Considering the fact that the side frame belongs to the class of structures of responsible purpose and is subject to multi-stage NDT, it is advisable to use machine vision technology as a visual-optical NDT system. Neural network algorithms form the basis of the software components of the visual-optical NDT system and automate the process of recognising surface defects. This approach contributes to the timely detection and elimination of defects, which leads to increased durability of the structure. 10 Ref., 3 Tabl., 9 Fig.
Keywords: critical structures, robotic welding, visual-optical non-destructive testing, defect detection, neural network, robotic system, machine vision


Received: 05.12.2025
Received in revised form: 25.03.2026
Accepted: 10.04.2026
Posted online 11.04.2026

References

1. Lobanov, L.M., Makhnenko, O.V., Pavlovsky, V.I., Saprykina, G.Y., Pustovoy, A.D. (2022) Advantages of an all-welded freight car bogie side frame. Nauka ta progress transportu, 3-4, 55–69. [in Ukrainian]. DOI: https://doi.org/10.15802/stp2022/276541
2. Lobanov, L.M., Makhnenko, O.V., Knysh, V.V., Solovej, S.A., Pavlovskyi, V.I. (2020) Development of welded structure of side frame of freight car truck of increased reliability. The Paton Welding J., 3, 13–18. DOI: https://doi.org/10.37434/tpwj2020.03.02
3. FANUC Robotics America Corporation. (2013) FANUC Robotics SYSTEM R-30iB ArcTool Setup and Operations Manual, Document ID: MAROBAR8203131E, REV A, Version 8.20 series.
4. Gonzalez, R.C., Woods, R.E. (2018) Digital Image Processing. 4th Edition, Pearson Education, New York.
5. Terven, J., Córdova-Esparza, D-M., Romero-González, J-A. (2023) A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS. Machine Learning and Knowledge Extraction, 5(4), 1680–1716. DOI: https://doi.org/10.3390/make5040083
6. Katsamenis, I., Karolou, E., Davradou, A. et al. (2022) Tra-Con: A novel dataset for real-time traffic cones detection using deep learning. Novel & Intelligent Digital Systems: Proceedings of the 2nd International Conference (NiDS 2022), 382–391. DOI: https://doi.org/10.1007/978-3-031-17601-2_37
7. Koay, H.V., Chuah, J H., Chow, C.-O., Chang, Y.-L., Yong, K.K. (2021) YOLO-RTUAV: Towards real-time vehicle detection through aerial images with low-cost edge devices. Remote Sensing, 13(21), 4196. DOI: https://doi.org/10.3390/rs13214196
8. Guan, S., Lei, M., Lu, H. (2020) A steel surface defect recognition algorithm based on improved deep learning network model using feature visualization and quality evaluation. IEEE Access, 8, 49885–49895. DOI: https://doi.org/10.1109/ACCESS.2020.2979755
9. Qike Wu (2024) NEU-DET. IEEE Dataport. DOI: https://doi.org/10.21227/j84r-f770
10. Hossin, M., Sulaiman, M.N. (2015) A review on evaluation metrics for data classification evaluations. International J. of Data Mining & Knowledge Management Process, 5, 1–11. DOI: http://dx.doi.org/10.5121/ijdkp.2015.5201

---

Suggested Citation

A.S. Novodranov, V.O. Koliada (2026) Investigation of the effectiveness of neural networks in welding of critical structures for reliability enhancement. Avtom. Zvaryuvannya, 02, 47-56.