Machine Learning-Assisted Efficient Design of Mg–Gd–Y Based System Alloys

With the rapid development of machine learning technology, its application in materials science is gradually becoming an important tool for mechanical property prediction and alloy design. In this paper, a machine learning based multi-objective optimization method is proposed to predict and optimize the yield strength (YS), ultimate tensile strength (UTS) and elongation (EL) of Mg–Gd–Y system alloys. Various advanced algorithms were used to construct efficient prediction models for YS, UTS, and EL, and the hyperparameters were tuned by a Bayesian optimization algorithm to improve the prediction accuracy. Subsequently, an innovative use of genetic algorithm (NAGA-III) was implemented for the multi-objective co-optimization of YS, UTS and EL to obtain the optimal solution for the alloy properties. On this basis, Shapley Additive Explanations interpretable analysis method was applied to dig deeper into the non-linear relationship between alloy composition and properties as well as the interactions of various factors, revealing the key influencing factors in alloy design. The experimental results show that the proposed method can effectively improve the accuracy of alloy property prediction and provide theoretical guidance and practical basis for the multi-objective design of Mg–Gd–Y system alloys.

Graphical Abstract