A normal weld recognition method for time-of-flight diffraction detection based on generative adversarial network

Hongquan Jiang (), Deyan Yang, Zelin Zhi, Qiangzheng Jing, Jianmin Gao, Chenyue Tao and Zhixiang Cheng
Additional contact information
Hongquan Jiang: Xi’an Jiaotong University
Deyan Yang: Xi’an Jiaotong University
Zelin Zhi: Xi’an Jiaotong University
Qiangzheng Jing: Xi’an Special Equipment Inspection Institute
Jianmin Gao: Xi’an Jiaotong University
Chenyue Tao: Xi’an Jiaotong University
Zhixiang Cheng: Xi’an Jiaotong University

Journal of Intelligent Manufacturing, 2024, vol. 35, issue 1, No 13, 217-233

Abstract: Abstract Time-of-flight diffraction (TOFD) has become a widely used nondestructive testing (NDT) technique, owing to its wide coverage, fast detection speeds, and high defect detection rates. However, compared with nondestructive radiographic testing images, TOFD image analysis requires more technicians and more difficult defect analysis. Owing to the improvements in weld manufacturing quality, there are fewer welds with defects; consequently, a large number of TOFD images have no defect information. The TOFD image analysis of normal welds occupies a lot of time in the weld evaluation process that easily leads to problems of missed and false detections and reduces the efficiency of overall weld evaluation. To solve these problems, a TOFD image reconstruction model based on the generative adversarial network (GAN) and a normal weld recognition method are proposed. First, combined with the TOFD image characteristics, an image-wave feature fusion (IWFF) module based on depth-separable convolution is designed, which integrates and analyzes the TOFD image and wave features, and an IWFF–GAN model is developed. Second, to improve the accuracy of normal weld recognition, a method for denoising the reconstructed error-feature map based on the total variation model is proposed. Finally, the proposed method is verified using the TOFD images of large-scale spherical pressure-tank welds. The results show that the method accurately distinguishes between the normal and abnormal welds, exhibiting a higher normal weld recognition accuracy. The area under the receiver operating characteristic curve is 0.9903.

Keywords: Time-of-flight diffraction; Normal weld recognition; Generative adversarial network; Image reconstruction (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s10845-022-02041-9 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:joinma:v:35:y:2024:i:1:d:10.1007_s10845-022-02041-9

Ordering information: This journal article can be ordered from
http://www.springer.com/journal/10845

DOI: 10.1007/s10845-022-02041-9

Access Statistics for this article

Journal of Intelligent Manufacturing is currently edited by Andrew Kusiak

More articles in Journal of Intelligent Manufacturing from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().