Abstract

Info

Title:: Rsearch on Semi-solid Process of High-strength Mg-Zn-Y-Zr Alloy Based on Machine Learning

Author(s):: Zeng Qi; Wang Shaoyang; Zhang Yingbo; Zhu Kai; Hu Yunfeng Author; 1. Chengdu Aircraft Industry(Group) Co. Ltd., Chengdu 610092, China 2. School of Materials Science and Engineering, Southwest Jiaotong University, Chengdu, 610031, Chin

Keywords:: Mg-Zn-Y-Zr alloy; machine learning; semi-solid isothermal treatment; quasicrystal; hot extrusion

Abstract:: In this paper, an improved Efficient Global Optimization (EGO) algorithm based on the Kriging model is adopted and applied to optimize the semi-solid isothermal treatment parameters of Mg-Zn-Y-Zr alloys containing quasicrystals. Firstly, the Mg-1.2Zn-0.2Y-0.15Zr alloy containing quasicrystalline phases was designed, cast and subjected to primary hot extrusion; then the sample data set and Kriging surrogate model between the semi-solid isothermal treatment parameters of the alloy and its mechanical properties were established; finally, by combining machine learning and the semi-solid isothermal treatment + hot extrusion composite processing technology, a significant optimization of the comprehensive mechanical properties of the alloy was achieved. Through a total of four iterative optimizations, the maximum tensile strength of 427.4 ± 2.4 MPa was obtained, and its semi-solid process parameters were 574 ℃ + 60 min. Compared with its primary extrusion state (285 ± 1.5 MPa), its tensile strength increased by 50%. It is proved that the machine learning method is feasible and efficient for optimizing the semi-solid isothermal treatment parameters and enhancing the strength of Mg-Zn-Y-Zr alloys.