Machine learning assisted multi-objective optimization for materials processing parameters: A case study in Mg alloy

Abstract Optimizing two or more targeted properties simultaneously is demanded in developing new materials, however, the combinatorial possibilities of chemical compositions and processing parameters are too large for an Edisonian approach to be practical. In the present study, a machine learning assisted strategy is formulated to iteratively recommend the next experiment to accomplish the multi-objective optimization in an accelerated manner. The efficacy of the strategy is demonstrated by optimizing the two step aging treatment parameters with the aim of enhancing the strength and ductility of as-cast ZE62 (Mg-6 wt.% Zn-2 wt.% RE) Mg alloy. The strength and ductility of the alloy are increased by 27% and 13.5% respectively with only four new successive experiments. The short-range homogenization is considered as the reason to enhance both properties. This strategy offers a recipe to solve the multi-objective optimization problems in designing composition and processing parameters of materials.