Skip to main content
SpringerLink
Log in

WeldNet: a lightweight deep learning model for welding defect recognition

  • Research Paper
  • Published:
Welding in the World Aims and scope Submit manuscript

Abstract

Weld defect detection is an important task in the welding process. Although there are many excellent weld defect detection models, there is still much room for improvement in stability and accuracy. In this study, a lightweight deep learning model called WeldNet is proposed to improve the existing weld defect recognition network for its poor generalization performance, overfitting, and large memory occupation, using a design with a small number of parameters but with better performance. We also proposed an ensemble-distillation strategy in the training process, which effectively improved the accuracy rate and proposed an improved model ensemble scheme. The experimental results show that the final designed WeldNet model performs well in detecting weld defects and achieves state-of-the-art performance. Its number of parameters is only 26.8% of that of ResNet18, but the accuracy is 8.9% higher, while achieving a 24.2 ms inference time on CPU to meet the demand of real-time operation. The study is of guiding significance for solving practical problems in weld defect detection, and provides new ideas for the application of deep learning in industry. The code used in this article is available at https://github.com/Wanglaoban3/WeldNet.git.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Algorithm 1
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10

Similar content being viewed by others

Data availability

Data will be made available on request.

Abbreviations

TIG:

Tungsten inert gas

CNN:

Convolutional neural network

SVM:

Support vector machine

CPU:

Central processing unit

GPU:

Graph processing unit

FCN:

Fully connected network

HDR:

High dynamic range

References

  1. Allen-Zhu Z, Li Y (2020) Towards understanding ensemble, knowledge distillation and self-distillation in deep learning. arXiv:2012.09816 (2020)

  2. Bacioiu D, Melton G, Papaelias M, Shaw R (2019) Automated defect classification of aluminium 5083 TIG welding using HDR camera and neural networks. J Manufac Process 45:603–613. https://doi.org/10.1016/j.jmapro.2019.07.020

    Article  Google Scholar 

  3. Bacioiu D, Melton G, Papaelias M, Shaw R (2019) Automated defect classification of SS304 TIG welding process using visible spectrum camera and machine learning. NDT & E International 107. https://doi.org/10.1016/j.ndteint.2019.102139

  4. Chen Z, Gao X (2014) Detection of weld pool width using infrared imaging during high-power fiber laser welding of type 304 austenitic stainless steel. Int J Adv Manufac Technol 74(9–12):1247–1254. https://doi.org/10.1007/s00170-014-6081-3

    Article  Google Scholar 

  5. Chiaranai S, Pitakaso R, Sethanan K, Kosacka-Olejnik M, Srichok T, Chokanat P (2023) Ensemble deep learning ultimate tensile strength classification model for weld seam of asymmetric friction stir welding. Processes 11(2):434

    Article  CAS  Google Scholar 

  6. Du D, Hou R, Shao J, Wang L, Chang B (2008) Real-time Xray image processing based on information fusion for weld defects detection. In: 17th world conference on nondestructive testing, Shanghai, China

  7. Ericsson M (2003) Influence of welding speed on the fatigue of friction stir welds, and comparison with MIG and TIG. Int J Fatigue 25(12):1379–1387. https://doi.org/10.1016/s0142-1123(03)00059-8

    Article  CAS  Google Scholar 

  8. Gao P, Wang C, Li Y, Cong Z (2015) Electromagnetic and eddy current NDT in weld inspection: a review. Insight-Non-Destructive Testing and Condition Monitoring 57(6):337–345

    Article  Google Scholar 

  9. Gao X, Zhang Y (2015) Monitoring of welding status by molten pool morphology during high-power disk laser welding. Optik - Int J Light Electron Optics 126(19):1797–1802. https://doi.org/10.1016/j.ijleo.2015.04.060

    Article  CAS  Google Scholar 

  10. Golodov VA, Maltseva AA (2022) Approach to weld segmentation and defect classification in radiographic images of pipe welds. NDT & E Int 127. https://doi.org/10.1016/j.ndteint.2021.102597

  11. He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 770–778

  12. Hinton G, Vinyals O, Dean J (2015) Distilling the knowledge in a neural network. arXiv:1503.02531

  13. Hou W, Zhang D, Wei Y, Guo J, Zhang X (2020) Review on computer aided weld defect detection from radiography images. Appl Sci 10(5). https://doi.org/10.3390/app10051878

  14. Howard AG, Zhu M, Chen B, Kalenichenko D, Wang W, Weyand T, Andreetto M, Adam HJapa (2017) MobileNets: efficient convolutional neural networks for mobile vision applications. arXiv:1704.04861

  15. Huang G, Li Y, Pleiss G, Liu Z, Hopcroft JE, Weinberger KQ (2017) Snapshot ensembles: train 1, get m for free. arXiv:1704.00109

  16. Huang J, Zhang Z, Qin R, Yu Y, Li Y, Wen G, Cheng W, Chen X (2023) Residual swin transformer-based weld crack leakage monitoring of pressure pipeline. Welding in the World pp 1–13

  17. Huang L, Liao C, Song X, Chen T, Zhang X, Deng Z (2020) Research on detection mechanism of weld defects of carbon steel plate based on orthogonal axial eddy current probe. Sensors (Basel) 20(19). https://doi.org/10.3390/s20195515. https://www.ncbi.nlm.nih.gov/pubmed/32993112. Huang, Linnan Liao, Chunhui Song, Xiaochun Chen, Tao Zhang, Xu Deng, Zhiyang eng 51807052/National Natural Science Foundation of China/ Switzerland Sensors (Basel). 2020 Sep 26;20(19):5515

  18. Krizhevsky A, Sutskever I, Hinton GE (2017) ImageNet classification with deep convolutional neural networks. Commun ACM 60(6):84–90

    Article  Google Scholar 

  19. Le Cun Y, Bottou L, Bengio Y (1997) Reading checks with multilayer graph transformer networks. In: 1997 IEEE International conference on acoustics, speech, and signal processing, IEEE 1:151–154

  20. Li H, Li L, Chen X, Zhou Y, Li Z, Zhao Z (2024) Addressing the inspection selection challenges of in-service pipeline girth weld using ensemble tree models. Eng Failure Anal 156:107852

    Article  Google Scholar 

  21. Li Z, Chen H, Ma X, Chen H, Ma Z (2022) Triple pseudo-siamese network with hybrid attention mechanism for welding defect detection. Mater & Design 217. https://doi.org/10.1016/j.matdes.2022.110645

  22. Ma G, Yu L, Yuan H, Xiao W, He Y (2021) A vision-based method for lap weld defects monitoring of galvanized steel sheets using convolutional neural network. J Manufac Process 64:130–139. https://doi.org/10.1016/j.jmapro.2020.12.067

    Article  Google Scholar 

  23. Mackwood AP, Crafer RC (2005) Thermal modelling of laser welding and related processes: a literature review. Optics & Laser Technol 37(2):99–115. https://doi.org/10.1016/j.optlastec.2004.02.017

    Article  Google Scholar 

  24. Madhvacharyula AS, Pavan AVS, Gorthi S, Chitral S, Venkaiah N, Kiran DV (2022) In situ detection of welding defects: a review. Welding in the World 66(4):611–628. https://doi.org/10.1007/s40194-021-01229-6

    Article  Google Scholar 

  25. Rumelhart DE, Hinton GE, Williams RJ (1986) Learning representations by back-propagating errors. Nature 323(6088):533–536

    Article  Google Scholar 

  26. Say D, Zidi S, Qaisar SM, Krichen M (2023) Automated categorization of multiclass welding defects using the X-ray image augmentation and convolutional neural network. Sensors 23(14):6422

  27. Silva LC, Simas Filho EF, Albuquerque MCS, Silva IC, Farias CTT (2020) Segmented analysis of time-of-flight diffraction ultrasound for flaw detection in welded steel plates using extreme learning machines. Ultrasonics 102:106057.https://doi.org/10.1016/j.ultras.2019.106057. https://www.ncbi.nlm.nih.gov/pubmed/31952796. Silva, Lucas C Simas Filho, Eduardo F Albuquerque, Maria C S Silva, Ivan C Farias, Claudia T T eng Netherlands Ultrasonics. 2020 Mar;102:106057. Epub 2019 Dec 11

  28. Simonyan K, Zisserman A (2014) Very deep convolutional networks for large-scale image recognition. arXiv:1409.1556

  29. Tian Y, Wang Y, Krishnan D, Tenenbaum JB, Isola P (2020) Rethinking few-shot image classification: a good embedding is all you need? In: Computer Vision-ECCV 2020: 16th European Conference, Glasgow, UK, August 23-28, 2020, Proceedings, Part XIV 16, Springer, pp 266–282

  30. Tyystjärvi T, Virkkunen I, Fridolf P, Rosell A, Barsoum Z (2022) Automated defect detection in digital radiography of aerospace welds using deep learning. Welding in the World 66(4):643–671

    Article  Google Scholar 

  31. Vilar R, Zapata J, Ruiz R (2009) An automatic system of classification of weld defects in radiographic images. NDT & E Int 42(5):467–476. https://doi.org/10.1016/j.ndteint.2009.02.004

    Article  CAS  Google Scholar 

  32. Wang X, Kondratyuk D, Christiansen E, Kitani KM, Alon Y, Eban E (2020) Wisdom of committees: an overlooked approach to faster and more accurate models. arXiv:2012.01988

  33. Xia C, Pan Z, Fei Z, Zhang S, Li H (2020) Vision based defects detection for keyhole TIG welding using deep learning with visual explanation. J Manufac Process 56:845–855. https://doi.org/10.1016/j.jmapro.2020.05.033

    Article  Google Scholar 

  34. Yan J, Gao M, Zeng X (2010) Study on microstructure and mechanical properties of 304 stainless steel joints by TIG, laser and laser-TIG hybrid welding. Optics Lasers Eng 48(4):512–517. https://doi.org/10.1016/j.optlaseng.2009.08.009

  35. Yu R, Kershaw J, Wang P, Zhang Y (2021) Real-time recognition of arc weld pool using image segmentation network. J Manufac Processes 72:159–167. https://doi.org/10.1016/j.jmapro.2021.10.019

Download references

Funding

This study is supported by Shandong Province Key Research and Development Program, No. 2020CXGC011201.

Author information

Authors and Affiliations

  1. College of New Energy, China University of Petroleum (East China), Qingdao, Shandong, 266580, People’s Republic of China

    Rongdi Wang, Hao Wang, Zhenhao He, Jianchao Zhu & Haiqiang Zuo

Authors
  1. Rongdi Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Hao Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Zhenhao He
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Jianchao Zhu
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Haiqiang Zuo
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Haiqiang Zuo.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Recommended for publication by Commission V - NDT and Quality Assurance of Welded Products.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Wang, R., Wang, H., He, Z. et al. WeldNet: a lightweight deep learning model for welding defect recognition. Weld World (2024). https://doi.org/10.1007/s40194-024-01759-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s40194-024-01759-9

Keywords

Search

Navigation