Forest of insights: Interpretable AI prediction of wear characteristics in WAAM-fabricated components

Accurately predicting wear behavior in WAAM-fabricated components is critical but traditionally costly and time-consuming. This study presents an interpretable AI approach to forecast region-specific wear in 316L stainless steel WAAM parts, using data from systematic tests across inter-/intra-layer regions, lubrication states, loads, and speeds. Among six tree-based models, Random Forest achieved the best performance for predicting Coefficient of Friction and Friction Force (R2=0.8997 and 0.9344, respectively). Feature importance and SHAP analysis revealed lubrication and load as dominant factors. The model’s performance was validated on unseen data, confirming its robustness and practical relevance. This approach offers a reliable path to reducing physical testing while providing deeper insights into tribological behavior.