Research on Welding Defect Classification Methods Based on Convolutional Neural Networks

Wei Tang; Yan Shi; Yinghua Liao; Qi Li; Yi Luo

doi:10.62051/ijcsit.v4n1.19

Authors

Wei Tang
Yan Shi
Yinghua Liao
Qi Li
Yi Luo

DOI:

https://doi.org/10.62051/ijcsit.v4n1.19

Keywords:

Convolutional neural networks, Multi-head attention mechanism, Welding defects, ResNet101

Abstract

To address the issues of low detection and identification rates and inaccurate classification of welding defects, this paper proposes a welding defect classification method based on convolutional neural networks. By modifying the ResNet101 backbone network and incorporating a multi-head attention mechanism (MHA), the method enhances the focus on regions of interest in welding defect images, thereby improving the ability to extract defect features. This approach reduces the chances of missed detections and false positives. By analyzing the effects of different optimizers and dynamic factors on recall (M) and precision (N), the optimal optimizer and dynamic factor are identified. The algorithm is validated on the JPEGWD dataset. Experimental results show that the recall (M) reaches 0.9634, representing a 4.88% improvement, and the precision (N) reaches 0.9875, representing a 3.81% improvement. The mean average precision (mAP) is 0.9817, reflecting a 5.98% improvement. These results effectively enhance the recognition rate of welding defects and reduce the probability of missed detections and false positives.

Downloads

Download data is not yet available.

References

[1] Li Yajuan, Wang Juan, Jiang Qinglei, et al. Analysis and Countermeasures of Welding Defects [M]. 3rd ed. Beijing: Chemical Industry Press, 2018: 3-81.

[2] Zhang Fulu, Wang Runzhi, Dong Ce, et al. Quantitative characterization of internal initiation fatigue crack growth rate [J]. Heat Treatment of Metals, 2019(S01): 611-614.

[3] Ye Zhilong. Research on Automatic Detection Methods of Welding Defects Based on Image Processing [D]. Nanjing University of Aeronautics and Astronautics, 2013.

[4] Wang Fei., Zhu Jianjiang. Application of Machine Vision in Welding Defect Detection Under Complex Environments [J]. Welding Technology, 2017, 46(05): 127-133.

[5] Liang Hongbao, Wang Lixun, Liu Lei. Current Status and Development of Non-Destructive Testing Technology for Pressure Vessels [J]. Petroleum Machinery, 2007, No.336(02): 54-57.

[6] Peng Shuaijun., Wang Kai, Wu Haifen, et al. Application of Radiographic Testing Technology in Dimensional Measurement [J]. Nondestructive Testing, 2007, No.272(11): 679-681.

[7] Liu Bin. Inspection of Surface Defects in Artificial Joint Metal Parts Using Penetrant Testing Technology [J]. Materials Engineering, 2009, No.312(05): 73-75.

[8] Kang Ziwen, Duan Yixiong, Wang Long. Research on Intelligent Magnetic Particle Testing Technology for Integrated Wheelsets [J]. Locomotive & Rolling Stock Technology, 2023, No.359(01): 47-50.

[9] Pan Haihong, Li Songting, Chen Lin, et al. Welding Seam Defect Recognition Method Based on Improved DG-MobileNet Model [J]. Modular Machine Tools and Automatic Manufacturing Techniques, 2023(08):127-130.

[10] Hu Jianfeng, Zhou Jiale, Wang Huifeng, et al. Research and Application of an Automatic Recognition Method for Electrofusion Welding Defects [J]. Journal of East China University of Science and Technology (Natural Science Edition), 2022, 48(04):526-533.

[11] Liu Mengxi, Ju Yongfeng, Gao Weixin, et al. Research on X-ray Welding Seam Defects Using Deep Convolutional Neural Networks [J]. Sensors and Microsystems, 2018, 37(05):37-39+43.

[12] Gu Jing, Wang Qiwen, Zhang Min, et al. Welding Seam Defect Detection and Recognition Based on DenseNet Network [J]. Sensors and Microsystems, 2020, 39(09):129-131.

[13] Qiao Li, Zhao Erdun, Liu Junjie, et al. Research on Workpiece Defect Detection Method Based on CNN [J]. Computer Science, 2017, 44(S2):238-243.

[14] Luo Shunhua, Wang Zhenlei, Wang Xin. Soft Measurement Modeling Based on Attention Mechanism with Multi-head CNN-LSTM [J]. Control Engineering, 2022, 29(10):1821-1828.

[15] Jin Guoqiang. Research on End-to-End Rolling Bearing Fault Diagnosis Algorithm Under Complex Working Conditions Based on Deep Learning [D]. University of Science and Technology of China, 2021.

[16] TAO SHEN, TIANYI ZHOU, GUODONG LONG, et al. Reinforced Self-Attention Network: a Hybrid of Hard and Soft Attention for Sequence Modeling[C]. //27th International Joint Conference on Artificial Intelligence and 23rd European Conference on Artificial Intelligence: IJCAI-ECAI 2018, Stockholm, Sweden, 13-19 July 2018, Volume 6 of 8: Curran Associates, Inc., 2018:4345-4352.

[17] JOAN SERRA, DIDAC SURIS, MARIUS MIRON, et al. Overcoming Catastrophic Forgetting with Hard Attention to the Task[C]. //35th International Conference on Machine Learning: ICML 2018, Stockholm, Sweden, 10-15 July 2018, volume 10 of 13: Curran Associates, Inc., 2018:7225-7242.

[18] Yao Zongwei, Yang Hongfei, Hu Jiyong, et al. Rail Surface Defect Detection Method Based on Machine Vision and Convolutional Neural Networks [J]. Journal of the China Railway Society, 2021, 43(04):101-107.