Machine learning guided high-throughput composition optimization and design of alloys
Yi Liu*, Wachen Zhao, Chen Zheng, Bin Xiao
Shanghai University, Shanghai, 200444, China
EXTENDED ABSTRACT: The traditional trail-and-error experiment development approaches rely heavily on the intuition and experience of researchers and are limited by the low efficiency of single sample experiment mode. To accelerate materials discovery, we developed a machine learning (ML) aided high-throughput experiment (HTE) approach to optimize the composition of alloys [1]. The aluminum (Al) 6061 has a wide range of composition consisting of multiple trace elements. It is important to understand the relationship between the enormous compositions and the mechanical properties of alloy 6061 in view of both scientific research and quality control in industry. In this work, we combined high-throughput experiments of materials synthesis and an active learning framework based on Bayesian optimization sampling strategy to develop an effective machine leaning model to describe the relationship between composition and hardness. We found the correlation between the content of trace alloying elements and hardness based on the machine learning prediction. Specifically, the error of hardness predicted by the Bayesian model is ~4.87 HV (6.12 %) using the 64 aluminum alloy samples after three rounds of iterations, much lower than the error 9.69 HV (15.4 %) predicted by the maximum expectation sampling method. This work provided efficient experimental strategies to improve the quality of product in industry by composition refinement and performance control of 6061 aluminum alloy within its standard nominal composition range.
REFERENCES [1] WC Zhao, …, Y Liu* et. al., Composition Refinement of 6061 Aluminum Alloy Using Active Machine Learning Model Based on Bayesian Optimization Sampling, Acta Matallurgica Sinica, 57(6), (2021) 797-810.
Prof. Yi LIU obtained his Ph. D. degree at Materials Science and Engineering at Institute of Metal Research in China in 1997. Then he has worked in the field of computational materials science at Nagoya University, Japan (1997-2002); Juelich Research Center, Germany (2002-2003); University of Western Ontario, Canada (2003-2005); California Institute of Technology, US (2006-2012). He is a professor at Materials Genome Institute and Department of Physics at Shanghai University (2015-present) after worked at the School of Materials Science and Engineering, the University of Shanghai for Science and Technology (2012-2015). His current research interests focus on the materials design for advanced alloys, combustion fuels, and nanomaterials by combining computation, machine learning, and high-throughput experiment approaches based on materials genome concept.