DBAN: Double Bias Adjustment Network for Domain Shift Defect Detection in Photovoltaic Intelligent Manufacturing

Developing reliable, generalized, and accurate defect detection technology for photovoltaic (PV) manufacturers is particularly critical with the demand for production line expansion. Existing technologies perform well when handling independently and identically distributed (IID) data. However, their performance significantly reduces when they encounter domain shift problems, including style and instance bias, prompted by production line expansion. In this paper, we propose a novel Dual Bias Adjustment Network (DBAN) to enhance the generalization and reliability of PV defect detection. Specifically, we construct a Global Style-Generalized Contrastive Learning (GSCL), which uses nonlinear transformation functions and contrastive learning strategies to enhance model adaptability for different style changes and global discriminative ability, effectively overcoming the style bias problem. We design a Test-Time Prototype Adjustment (TTPA) that employs graph methods and prototype learning to adjust feature representations accurately. TTPA enhances prediction reliability during testing via dynamic prototype repositories and memory mechanisms, effectively addressing instance deviations. We conduct comprehensive experiments proving that DBAN achieves optimal performance, surpassing other advanced algorithms. Moreover, GSCL and TTPA add little inference time to the model, making them suitable for practical industrial applications. Finally, we conduct extensive experiments on the public domain-shifted PV dataset ELES, where our model achieves state-of-the-art performance in Single-domain generalized object detection. Note to Practitioners—This work proposes a practical defect detection solution, DBAN, enabling PV manufacturers to maintain reliable quality control across multiple production lines. The model can be directly integrated into existing inspection systems by deploying DBAN software on a central processing server that receives EL images from production lines. Trained with historical defect data from one production line, the model can automatically inspect products from expanded lines and continuously update its detection capabilities using real-time test data without retraining. DBAN enables practitioners to monitor multiple lines through a unified interface while maintaining high-quality standards, thus reducing equipment and labor costs. The system continuously enhances detection accuracy under varying manufacturing conditions, making it especially valuable for manufacturers seeking efficient, scalable production while ensuring product quality.