Machine learning-driven multi-objective alloy selection framework for mechanical property criteria

Abstract

This study presents a new machine learning framework for multi-objective alloy selection, focusing on key mechanical properties such as tensile strength, elongation, hardness, and Charpy energy. Traditional tools are often limited in their ability to manage large datasets and the complex, non-linear relationships between alloy composition, process parameters, and mechanical properties. In contrast, machine learning models such as XGBoost, Fine-Tuned Stacking, and Ensemble methods provide a scalable solution, allowing for the simultaneous consideration of multiple mechanical property objectives. The models were trained on a comprehensive dataset of stainless steel alloys, filtering materials that meet predefined performance criteria. Among the models, the Ensemble approach achieved the best results, with a precision of 0.98 and recall of 0.93. The findings show that integrating machine learning into the alloy selection process has the potential to improve decision-making accuracy for practical engineering applications.