Accelerated optimal design of high-performance Mg alloys through integration of Bayesian optimization with active machine learning

This research investigates the accelerated design of high-performance magnesium (Mg) alloys by integrating Bayesian optimization with active machine learning. The dataset encompasses various alloy compositions and processing parameters, including Zn, Y, Y, Zr, Nd, and Gd concentrations, along with solution treatment, homogenization, extrusion, and aging conditions. Mechanical properties such as Ultimate Tensile Strength (UTS), Yield Strength (YS), and Elongation (EL) are also recorded. Key findings indicate that increased Zn content generally enhances UTS and YS, though at the expense of elongation. Bayesian optimization and machine learning techniques applied to this data enable efficient identification of optimal alloy compositions and processing parameters, significantly accelerating the development of high-performance Mg alloys.

本研究探索了将贝叶斯优化与主动机器学习相结合，以加速高性能镁（Mg）合金的设计工作。本数据集涵盖多种合金成分与工艺参数，包括锌（Zn）、钇（Y）、锆（Zr）、钕（Nd）以及钆（Gd）的含量，同时包含固溶处理、均质化处理、挤压成型与时效处理的相关工艺条件。此外，数据集还记录了极限抗拉强度（Ultimate Tensile Strength，以下简称UTS）、屈服强度（Yield Strength，以下简称YS）与断后伸长率（Elongation，以下简称EL）等力学性能指标。核心研究结果显示，提升锌含量通常可有效增强合金的极限抗拉强度与屈服强度，但会以牺牲断后伸长率为代价。基于本数据集应用贝叶斯优化与机器学习技术，能够高效筛选出最优合金成分与工艺参数，显著加速高性能镁合金的开发进程。