Optimisation and predictive modelling of hardness and impact energy in friction stir welded Ni-coated Al₂O₃ reinforced 7075 aluminium composites

The study investigates the influence of friction stir welding (FSW) parameters on the hardness and impact energy of AA-7075 composites reinforced with 6 wt% nickel-coated alumina particles. The influence of tool rotation speed, welding speed, axial force and tool tilt angle on the performance of the joint was systematically examined. Hardness and impact energy were chosen as the responses owing to their sensitivity to weld integrity and damage tolerance. Experiments were designed using Taguchi’s L27 orthogonal array. Consequently, multi-response optimisation was performed using Complex Proportional Assessment (COPRAS) to achieve balanced improvement in both responses. The optimal condition (6 kN, 60 mm/min, 600 rev min−1 and 1°) resulted in enhanced hardness and impact energy. To minimise the experimental dependence and support parameter prediction, machine learning (ML) models such as Decision Tree, Random Forest, Neural Networks and Gaussian Process were implemented and assessed with statistical metrics. The predicted results converged towards the COPRAS-derived optimal parametric set. Further, experimental validation yielded hardness of 143.41 ± 2.7 HV and impact energy of 4.7 ± 0.4 J, confirming reproducibility. The proposed optimisation–prediction framework provides an effective approach for designing high-performance welded joints in advanced aluminium matrix composites.