• • Optimal parameters (230 W, 14 mm·s⁻¹, 75% overlap) yield a hardened layer depth of 230 μm and fused layer depth of 66 μm, with prediction errors of 6.52% and 9.09%, respectively; these depths balance wear resistance and toughness, critical for industrial components like crankshafts and gears.
• • Fused layer microhardness reaches 940 ± 40 HV0.5 and hardened layer 630 ± 30 HV0.5, representing a 5.7× and 3.8× increase over the substrate (166 ± 15 HV0.5), directly enhancing resistance to abrasive wear and contact fatigue in ductile iron parts.
• • The BO-RF-XGBoost model outperforms standalone RF, XGBoost, and RF-XGBoost ensembles, reducing prediction error by up to 30% compared to baseline models, enabling reliable virtual process qualification and reducing trial-and-error costs.
• • Finite element model validation shows relative errors of 3.03% for fused layer depth and 9.13% for hardened layer depth, confirming the phase-transformation heat-transfer model's accuracy for simulating laser hardening without costly physical experiments.