A real spatial–temporal attention denoising network for nugget quality detection in resistance spot weld

Jie Zhou (), Zerui Xi (), Shilong Wang (), Bo Yang (), Youhong Zhang () and Yucheng Zhang ()
Additional contact information
Jie Zhou: Chongqing University
Zerui Xi: Chongqing University
Shilong Wang: Chongqing University
Bo Yang: Chongqing University
Youhong Zhang: Seres Group Co. Ltd.
Yucheng Zhang: Chongqing Seres New Energy Automobile Design Institute Co., Ltd.

Journal of Intelligent Manufacturing, 2024, vol. 35, issue 6, No 16, 2743-2764

Abstract: Abstract Resistance spot welding is an important process in the production of body-in-white. The quality of the welded nugget affects the safety performance of the whole vehicle. Currently, the quality of the welded nugget is mainly inspected manually, which is labor-intensive and inefficient. Therefore, this paper explores a new method to automatically and efficiently detect the quality of welded nugget by analyzing the vibration excitation response signal of the welded joint. In response to the characteristics of large volume, significant noise, and small discriminability of raw signals, we constructed a deep learning model named Real Spatial–temporal Attention Denoising Network (RSTADN), which consists of a denoising module, spatial–temporal attention modules, and multiple residual modules. The denoising module uses global absolute average pooling (GAAP) to maximize the retention of the original signal characteristics while aggregating global information from each channel. It generates appropriate soft thresholds to eliminate noise and enhance the feature recognition ability of the model. The spatial–temporal attention modules delve into the spatiotemporal correlation features of the signal from different perspectives, including real spatial scale, short-term dependency, and global temporal interaction, to enhance the model's feature extraction ability. Multiple residual modules further extract signal features to achieve precise alignment between features and nugget quality states. The experimental results indicate that the accuracy of RSTADN in the task of detecting the quality of welded nugget reaches as high as 94.35%, which is at least 1.31% higher than that of existing models.

Keywords: Resistance spot welding; Welded nugget quality; Deep learning; Attention mechanism; Vibration excitation response signal analysis (search for similar items in EconPapers)
Date: 2024
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s10845-023-02160-x 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:6:d:10.1007_s10845-023-02160-x

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

DOI: 10.1007/s10845-023-02160-x

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 ().