A Novel Method for Alloy Materials Design via Sample Data Transfer Machine Learning

Abstract

One of the challenges in material design is to rapidly develop new materials or improve the performance of materials by utilizing the data and knowledge of existing materials. Here, a novel method of alloy material design via data transfer learning is proposed to efficiently design new alloys using existing data. A new type of aluminum alloy with ultra strength and high toughness (so-called E2 alloy) previously developed by the authors is used as an example, and an optimal three-stage solution-aging treatment process (which is very difficult to design by trial-and-error method, T66R process for short) was rapidly designed based on 1053 pieces of heat treatment process data of AA7xxx series commercial high-strength aluminum alloy and a total of 29 groups of experimental data of E2 alloy. It overcomes the bottleneck of expensive theoretical calculations and more than millions of possible process combinations in the experimental trial-and-error space in the process design of complex alloy heat treatment. Compared with the common T6 process in the aluminum industry, the T66R process can increase the ultimate tensile strength and elongation of the E2 alloy simultaneously, from 715 ± 6 MPa and 8.4 ± 0.4% to 767 ± 6 MPa and 13.4 ± 0.5%, respectively. The results of microstructure characterization indicate that employing the optimal T66R process, the amount of micron-scale insoluble phases in the E2 alloy is sharply reduced, and the number of nano-precipitates is significantly increased, which leads to the simultaneous improvement of the strength and plasticity of the alloy. This study shows that transferring the existing alloy data is an effective method to design new alloys.