AZ91 Magnesium–alumina composites drilling with sustainable coolants: Surface roughness and axial thrust force using neural-guided genetic algorithms

AZ91 Magnesium (Mg) alloy reinforced with Alumina (Al 2 O 3 ) foam has attracted attention for lightweight structural and biomedical applications due to its high specific strength and corrosion adaptability. However, the drilling behavior of Mg systems containing porous ceramic reinforcements remains insufficiently characterized, particularly under sustainable cooling environments. The present study investigates axial thrust force and surface roughness during drilling of AZ91–Al 2 O 3 foam composites using a Taguchi L18 experimental design under vegetable oil, coconut oil, and liquid nitrogen cooling conditions. Experimental results indicate that axial thrust force varied from 104.3 N to 249 N across the investigated parameter space, while surface roughness ranged from 1.3 μm to 2.7 μm. Lower feed rates (0.2 mm/rev) and higher cutting speeds (≈180 m/min) were observed to reduce thrust force and improve surface integrity. A feed-forward artificial neural network (ANN) model achieved coefficients of determination of 0.98 for thrust force and 0.97 for surface roughness prediction, indicating strong nonlinear mapping capability within the defined experimental window. Multi-objective optimization using Non-dominated Sorting Genetic Algorithm (NSGA-II) identified an optimal parameter combination of 180 m/min cutting speed, 0.2 mm/rev feed rate, high-speed steel tool, and coconut oil coolant. Confirmation testing yielded 109.2 N axial thrust force and 1.28 μm surface roughness, with deviations of 3.2% and 1.7%, respectively, from predicted values. The proposed ANN–NSGA-II framework establishes a quantified and interpretable optimization strategy for sustainable drilling of reinforced Mg composites, enabling balanced reduction of mechanical load and surface irregularity under bio-lubricated conditions.