Automatic Weld Bead Discontinuities Detection based on Dye Penetrant Test by Deep Learning
DOI:
https://doi.org/10.53840/myjict10-1-141Keywords:
Welding, Deep Learning, SqueezeNet Model, Die Penetrant TestAbstract
Dye Penetrant Test is widely used in manufacturing industries, solely inspected by human eyes. The test using the human eye to detect the dye trap in weld surface cavity discontinuities – crack and porosity. However, the test is prone to the human factor effects and this led to erroneous results, hence automation is desired. This paper aims to propose a system by Deep Learning approach using SqueezeNet Framework based on the test. A dataset was obtained from a set of weld bead soft steel plates welded by arc welding process covering 258 RGB digital images in three categories – normal, crack, and porosity. The result through the proposed system is excellent by 100% precision, 87.5% recall, and 92% accuracy. Besides, it is statically good with F-score at 93.33% and found statistically significant. This result is uplifting, hence by this data the proposed system is acceptable to replace human inspection.
Downloads
References
American Society for Testing and Materials. (August, 2023). Standard Practice for Liquid Penetrant Testing for General Industry (ASTM E165/E165M-23). ASTM International. https://www.astm.org
Beale, M. H., Hagan, M. T., & Demuth, H. B. (2020). Deep learning toolbox user’s guide. The Mathworks Inc.: Herborn, MA, USA.
Bengio, Y. (2009). Learning deep architectures for AI. Foundations and trends® in Machine Learning, 2(1), 1-127.
Bengio, Y. (2012). Practical recommendations for gradient-based training of deep architectures. In Neural Networks: Tricks of the Trade: Second Edition (pp. 437-478). Berlin, Heidelberg: Springer Berlin Heidelberg.
Charles, R. L., Johnson, T. L., & Fletcher, S. R. (2015). The use of job aids for visual inspection in manufacturing and maintenance. Procedia CIRP, 38, 90-93.
Dwivedi, D. K. (2022). Fundamentals of metal joining. Springer Singapore.
Endramawan, T., & Sifa, A. (2018, February). Non-destructive test dye penetrant and ultrasonic on welding SMAW Butt joint with acceptance criteria ASME standard. In IOP Conference Series: Materials Science and Engineering (Vol. 306, No. 1, p. 012122). IOP Publishing.
Eisenmann, D. (2018). Liquid Penetrant Inspection. In Anderson, K., Weritz, J., & Kaufman, J. G. (Ed.). ASM Handbook, Volume 2A: Aluminum Science and Technology (pp. 46-59). ASM International.
Glebov, V., & Lashmanov, O. U. (2015, June). Modelling of microcracks image treated with fluorescent dye. In Modeling Aspects in Optical Metrology V (Vol. 9526, pp. 316-321). SPIE.
GÜLTEKİN, E., ÇELİK, H. İ., DÜLGER, L. C., SÜNBÜL, H. İ., & Harun, K. A. N. İ. (2019). Image Processing Applications on Yarn Characteristics and Fault Inspection. Tekstil ve Mühendis, 26(116), 340-345.
Karatay, M., FZ, B. M., Guldu, K., & Harmansah, C. (September 2-4, 2017). Investigation of Usability of Fluorescent and Radiolabeled Liquid Penetrant as a Non-Destructive Testing Agent. The 14th International Conference of the Slovenian Society for Non-Destructive Testing, Bernardin, Slovenia.
Koshti, A. M. (2018, March). Assessment of dye penetrant crack detectability in external corners using similarity analysis. In Nondestructive Characterization and Monitoring of Advanced Materials, Aerospace, Civil Infrastructure, and Transportation XII (Vol. 10599, pp. 322-331). SPIE.
Rodríguez-Gonzálvez, P., Rodríguez-Martín, M., Ramos, L. F., & González-Aguilera, D. (2017). 3D reconstruction methods and quality assessment for visual inspection of welds. Automation in construction, 79, 49-58.
Shipway, N. J. (2019). Automated defect detection for fluorescent penetrant inspection using machine learning (Doctoral dissertation, Imperial College London). https://www.imperial.ac.uk/
Shipway, N. J., Huthwaite, P., Lowe, M. J. S., & Barden, T. J. (2021). Using ResNets to perform automated defect detection for Fluorescent Penetrant Inspection. NDT & E International, 119, 102400.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Malaysian Journal of Information and Communication Technology (MyJICT)
This work is licensed under a Creative Commons Attribution 4.0 International License.
