化学学报 ›› 2023, Vol. 81 ›› Issue (2): 158-174.DOI: 10.6023/A22110446 上一篇    下一篇

综述

机器学习在新材料筛选方面的应用进展

戚兴怡, 胡耀峰, 王若愚, 杨雅清, 赵宇飞*()   

  1. 北京化工大学化学学院 化工资源有效利用国家重点实验室 北京 100029
  • 投稿日期:2022-11-01 发布日期:2022-12-23
  • 通讯作者: 赵宇飞
  • 作者简介:

    戚兴怡, 北京化工大学化学学院化学专业在读研究生. 2022年6月于北京化工大学化学学院应用化学专业获得学士学位.

    胡耀峰, 北京化工大学化学学院化学专业在读本科生.

    赵宇飞, 北京化工大学化学学院教授, 博士生导师. 入选中国科协“青年人才托举工程”计划, 2019年度国家自然科学基金优秀青年科学基金获得者. 工作围绕水滑石基二维插层材料的可控合成及精细结构表征, 面向高值精细化学品的光/电催化合成. 近年来以第一/通讯作者在Chem. Soc. Rev.、J. Am. Chem. Soc.、Angew. Chem.、Adv. Mater.、Chem、Joule、Ind. Eng. Chem. Res.等期刊上发表SCI收录论文30余篇; 累计SCI他引4500余次; 5篇入选ESI高被引论文, 已授权国家发明专利19项.

    同等贡献
  • 基金资助:
    国家自然科学基金(21922801); 国家自然科学基金(22090032); 国家自然科学基金(22090030); 北京自然科学基金(2202036)

Recent Advance of Machine Learning in Selecting New Materials

Xingyi Qi, Yaofeng Hu, Ruoyu Wang, Yaqing Yang, Yufei Zhao()   

  1. State Key Laboratory of Chemical Resource Engineering, College of Chemistry, Beijing University of Chemical Technology, Beijing 100029
  • Received:2022-11-01 Published:2022-12-23
  • Contact: Yufei Zhao
  • About author:
    Contributed equally to this work
  • Supported by:
    National Natural Science Foundation of China(21922801); National Natural Science Foundation of China(22090032); National Natural Science Foundation of China(22090030); Natural Science Foundation of Beijing(2202036)

新材料产业是许多相关领域技术变革的基础, 也是新能源、航空航天、电子信息等高新技术产业发展的先导. 传统研发手段由于成本高、效率低、商业化周期长等不利因素无法满足现代社会的发展需求. 近年来大数据与人工智能不断深入结合, 以数据驱动为核心的机器学习在新材料设计、筛选以及性能预测等方面取得巨大进展, 极大促进了新材料的研发与应用. 本综述总结了机器学习的基本过程及其在材料科学中常用的算法和相关材料数据库, 重点介绍了机器学习在不同功能上的应用以及在催化剂材料、锂离子电池、半导体材料和合金材料等领域的性能预测和材料开发中的最新进展, 并对其下一步在新材料应用方面提出展望.

关键词: 机器学习, 材料科学, 材料基因组, 高通量计算

The new material industry is the foundation of technological change in many related fields, and also the forerunner of the development of new energy, aerospace, electronic information and other high-tech industries. Traditional means cannot meet the development needs of modern society because of disadvantages such as high cost, low efficiency and long commercial cycle. In recent years, with the application of big data combined with artificial intelligence in a deeper degree, data-driven machine learning has made great progress in the design, screening and performance prediction of new materials, which has greatly promoted the development and application of new materials. In this review, the basic process of machine learning, the algorithms commonly used in materials science and the relevant materials database are summarized. This review focuses on the application of machine learning in different functions, as well as the performance prediction in the fields of catalyst materials, lithium-ion batteries, semiconductor materials and alloy materials, presenting the latest progress in materials development. Finally, machine learning in the application of new materials are analyzed and prospected.

Key words: machine learning, materials science, material genome, high throughput computing