An Integrated Framework for the Weldability of Metallic Materials

Weldability is a multiparametric property governed by interacting metallurgical, production, and design-related factors that cannot be adequately represented by a single material descriptor. This study proposes an integrated quantitative framework in which material- and process-specific risk factors are normalized and aggregated into an Equivalent Risk Factor, which is subsequently transformed via a five-parameter logistic function into a continuous Weldability Number (WN; 0–1). The framework is general in its structure and evaluation methodology, while the risk factor matrices, weighting coefficients, thresholds, and model parameters require calibration for individual classes of metallic materials. Experimental validation was performed on 2 mm S235JR steel welded via laser beam welding, MAG, and TIG, yielding WN values of 0.885, 0.813, and 0.833, respectively, with good agreement between predicted and experimentally derived values. A YOLO-based machine vision module was additionally implemented for automated visual inspection of welding imperfections, demonstrating how AI-assisted quality control can be integrated into the proposed architecture. The present results validate the framework for the investigated steel conditions rather than establishing universal quantitative parameterization. Large material- and process-specific datasets, combined with machine-learning interpolation and parameter optimization, are proposed as the next step toward refining the risk matrices and developing a broadly applicable weldability assessment system.