Journal of Alloys and Compounds2022,Vol.90315.DOI:10.1016/j.jallcom.2022.163828

High strength aluminum alloys design via explainable artificial intelligence

Park S. Seo E. Kim H. Yadav B.N. Jung I.D. Kayani S.H. Sung H. Euh K. Park S.J.
Journal of Alloys and Compounds2022,Vol.90315.DOI:10.1016/j.jallcom.2022.163828

High strength aluminum alloys design via explainable artificial intelligence

Park S. 1Seo E. 2Kim H. 2Yadav B.N. 2Jung I.D. 2Kayani S.H. 1Sung H. 1Euh K. 3Park S.J.4
扫码查看

作者信息

  • 1. Department of Materials Engineering and Convergence Technology (Center for K-metals) Gyeongsang National University
  • 2. Department Mechanical Engineering Ulsan National Institute of Science and Technology
  • 3. Korea Institute of Materials Science
  • 4. Department of Mechanical Engineering Pohang University of Science and Technology
  • 折叠

Abstract

? 2022Here, we have approached to discover new aluminum (Al) alloys with the assistance of artificial intelligence (A.I.) for the enhanced mechanical property. A high prediction rate of 7xxx series Al alloy was achieved via the Bayesian hyperparameter optimization algorithm. With the guide of A.I.-based recommendation algorithm, new Al alloys were designed that had an excellent combination of strength and ductility with a yield strength (YS) of 712 MPa and elongation (EL) of 19%, exhibiting a homogeneous distribution of nanoscale precipitates hindering dislocation movement during deformation. Adding Mg and Cu was found to be the critical factor that decides the relative ratio of strength and EL. We also demonstrate an explainable A.I. (XAI) system that reveals the relationship between input and output parameters. Our A.I. assistant system can accelerate the search for high-strength Al alloys for both experts and non-experts in the field of Al alloy design.

Key words

7xxx aluminum alloys/A.I.-based recommendation algorithm/Alloy design/Deep neural networks/Explainable artificial intelligence/Hyperparameter tuning

引用本文复制引用

出版年

2022
Journal of Alloys and Compounds

Journal of Alloys and Compounds

EISCI
ISSN:0925-8388
被引量9
参考文献量90
段落导航相关论文