Abstract

The slitting of aluminum (Al) foil is a critical process in secondary battery manufacturing, where cut quality directly affects electrode uniformity and production yield. Although femtosecond (fs) laser processing has emerged as a promising approach for high-precision foil cutting, residual debris generated during material removal can degrade product quality and requires accurate process evaluation. In this study, a hybrid framework combining adaptive computer vision and eXtreme Gradient Boosting (XGBoost) was developed for automated debris quantification, quality classification, and process optimization of fs laser-processed Al foils. The image processing algorithm automatically detects debris boundaries from scanning electron microscopy images and extracts key geometrical descriptors, which are subsequently used as input features for XGBoost models. The developed framework successfully distinguished acceptable and defective processing conditions and accurately predicted residual debris sizes. Experimental validation under previously unseen processing conditions confirmed excellent agreement between predicted and measured debris sizes. By enabling automated and interpretable quality assessment, the proposed framework provides a scalable foundation for defect quantification and machine-learning-assisted process optimization in advanced laser manufacturing.

IPC Classification

Keywords