Experimental and machine learning investigation of dry sliding wear behavior in multiwall carbon nanotube reinforced stir-cast Al7475 alloy

The present study investigates the mechanical and dry-sliding tribological behaviour of MWCNT-reinforced Al7475 nanocomposites fabricated through stir casting. Microstructural characterization, hardness evaluation, surface roughness analysis, and pin-on-disc wear tests were performed across different reinforcement levels and operating conditions. A random forest machine-learning model enhanced with SHapley Additive exPlanations (SHAP) analysis was developed to interpret wear behaviour and quantify the influence of reinforcement, load, speed, and sliding distance. The results showed that heat treatment increased hardness by 15 %, with the 0.5 and 0.75 wt% composites achieving the highest values (165 HVN). The 0.75 wt% MWCNT composite exhibited the lowest specific wear rate (3.6 mm³ N⁻¹m⁻¹), a reduced coefficient of friction, and improved resistance to surface damage. SHAP analysis revealed that MWCNT content was the most influential factor governing wear, followed by applied load, sliding speed, and sliding distance. Gray relational analysis identified 0.75 wt% MWCNT, 10 N load, 1000 rpm, and 750 m as the optimal combination of parameters. Overall, the study confirms that controlled MWCNT addition significantly enhances wear resistance and surface integrity of Al7475 alloys, and interpretable machine-learning tools provide reliable predictive insights for tribological optimization.