Heat treatment-driven optimization of Al7075-AlN composites: A synergistic approach using experiments, statistics, and machine learning

This article gives insights of a mathematical model for the Hardness of Al7075-AlN MMCs. This study presents a comprehensive investigation into the fabrication, characterization, and predictive modelling of Al7075-Aluminium Nitride (AlN) metal matrix composites (MMCs). Al7075 alloy was reinforced with varying weight percentages of AlN (0–10 wt.%) through the stir casting process, followed by forging and T6 heat treatment involving solutionizing at 470 °C and artificial ageing between 2–10 hours. Different quenching media (air, water, and ice) were employed to evaluate the influence of cooling rates on composite properties. Experiments were designed using Taguchi’s L9 orthogonal array to systematically explore the effects of reinforcement content, ageing duration, and quenching media on hardness. Statistical analysis and regression modelling, performed using MINITAB R18, demonstrated a strong correlation between process parameters and hardness, with the developed model achieving an R2 value of 0.9601, indicating high predictive accuracy. Experimental validation confirmed the reliability of the model, as experimental hardness closely matched theoretical predictions with a variance of less than 5%. Microstructural analysis via SEM and phase identification through XRD confirmed uniform distribution of AlN reinforcement, effective grain refinement, and the formation of beneficial intermetallic phases. Furthermore, machine learning models, particularly LightGBM, outperformed others with an R2 of 0.9745, enhancing prediction accuracy for hardness outcomes. Feature importance analysis revealed that the quenching medium had the most significant effect, followed by reinforcement percentage and ageing time. Overall, the integration of experimental, statistical, and machine learning approaches offers a robust framework for optimizing the mechanical properties of Al7075-AlN composites.