AI summaryⓘ
The authors found that deep learning models for classifying welding quality can fail when used on different types of welding processes due to differences in how the welds are made. To fix this, they designed a method that helps the model adapt to new welding types without needing extra labeled data, using a technique called unsupervised domain adaptation combined with gradually expanding the training data. Their method works well on both familiar and new welding types, achieving much better accuracy than traditional approaches. They also showed that the model learns to recognize important features that work across welding types, reducing the need for costly manual labeling when switching to new welding machines.
supervised deep learningweld penetration classificationdomain shiftunsupervised domain adaptationgradual source domain expansionTIG weldinglaser weldingUMAP visualizationcross-process transfermodel generalization
Authors
Sen Li, Haichao Cui, Chendong Shao, Yaqi Wang, Xinhua Tang
Abstract
Supervised deep learning has been widely used for weld penetration state classification; however, its performance often degrades significantly under domain shift, such as when transferring models between welding processes with distinct physical mechanisms:for instance, from arc-dominated tungsten inert gas (TIG) welding to keyhole-based laser welding. To overcome this limitation, we propose an unsupervised domain adaptation (UDA) framework integrated with a gradual source domain expansion (GSDE) strategy. Evaluated on dedicated TIG and laser welding datasets, our approach achieves high accuracy in both same-process and cross-process transfer tasks. Specifically, it attains average accuracies of 90.65% on TIGFH and 90.72% on LSPS in same-process settings, surpassing a supervised baseline by 35.83% and 38.87%, respectively. More notably, in cross-process scenarios, it reaches 80.48% for TIG to Laser and 81.13% for Laser to TIG, improving upon the baseline by 43.39% and 43.40%. UMAP visualizations verify that the model learns domain-invariant features while maintaining discriminative class boundaries. This method considerably lowers the relabeling cost for new welding processes and enhances the versatility of intelligent monitoring across different welding systems.