Harnessing machine learning approach for hardness optimization of Al-Si alloy composites reinforced with coconut shell ash

Abstract Purpose . The purpose of this study was to utilize Machine Learning (ML) to optimize the hardness of aluminum-silicon (Al-Si) alloy composites reinforced with Coconut Shell Ash (CSA). Specifically, the study focused on leveraging ML techniques to accurately predict material properties and streamline optimization. Also, the study intended to optimize the composition and process parameters simultaneously through the ML approach. Methodology . The study used Minitab’s Automated Machine Learning (AutoML), specifically the TreeNet model, to develop a predictive model for hardness. The input parameters included Al-Si alloy content, CSA content, melting temperature, and stirring speed. The model was trained and validated using experimental data, and hyperparameter tuning was performed to improve accuracy. The optimal settings were then experimentally verified to assess the model’s reliability. Findings . The optimal composition for maximizing hardness was 90 wt% Al-Si alloy and 10 wt% CSA, with a melting temperature of 800 °C and a stirring speed of 800 rpm. The experimental results validated the model’s predictions, with a hardness value of 70.9 BHN. The study demonstrated that CSA can be an effective, eco-friendly reinforcement for Al-Si composites, enhancing mechanical properties and promoting sustainability. Practical Implications . The findings have significant implications for industries such as automotive, aerospace, and defense, where lightweight, high-strength materials are critical. The ML-based approach used in this study can reduce the need for extensive experimental testing, offering a practical and efficient method for optimizing composite materials. Using CSA as a reinforcement also contributes to sustainable manufacturing by utilizing agricultural waste. Originality . This study presents an innovative approach by integrating ML into optimizing metal matrix composites, specifically using an eco-friendly reinforcement material. The article offers a novel contribution to materials science and sustainable engineering through a novel and structured step-by-step AutoML approach.