基于计算机技术的材料研发方法概述  

A Review of Material Research and Development Methods Based on Computer Technology

作　　者：郭佳龙 王宗国[1,2] 王彦棡 赵旭山[1] 宿彦京 刘志威[1,2] GUO Jialong;WANG Zongguo;WANG Yangang;ZHAO Xushan;SU Yanjing;LIU Zhiwei(Computer Network Information Center,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100049,China;Institute for Advanced Materials and Technology,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]中国科学院计算机网络信息中心,北京100190 [2]中国科学院大学,北京100049 [3]北京科技大学新材料技术研究院,北京100083

出　　处：《数据与计算发展前沿》2021年第2期120-132,共13页Frontiers of Data & Computing

基　　金：国家自然科学基金青年基金(51802312和51701208);中国科学院信息化专项(No.XXH13506-410);中国科学院前沿科学重点研究计划(No.ZDBS-LY-7025)。

摘　　要：【目的】随着大数据时代的到来和材料基因组计划的提出,材料的研发模式开始由“试错法”向科学研究第四范式转变。本文主要针对新材料研发模式中涉及的计算机技术进行概述。【方法】作者跟踪调研了近年来材料计算科学和数据驱动材料研发的文献资料,对计算机技术在材料研发中的应用进行了总结和讨论。【结果】以材料计算科学和“数据+人工智能”驱动为基础的创新模式,可以显著提升材料研究效率,从多维度、新角度探索材料多参数或性能之间的关联关系。【结论】计算机技术在材料研发中的广泛应用对加快新材料研发、加深对材料的认知以及关键技术的突破具有重要意义。[Objective]With the advent of the era of big data and the proposal of the Materials Genome Initiative,the research and development mode of materials has experienced a fundamental transformation from"trial and error"to the fourth paradigm of material science.This paper aims to summarize the computer technologies involved in this new materials research mode.[Methods]Based on the literatures on material computation and data-driven techniques in recent years,the paper summarizes and discusses the application of computer technologies in material research and development.[Results]The innovation mode marked by“data+artificial intelligence”and computational material science can improve the efficiency of material research significantly,and can explore the relationship between various material parameters and performance from multi-dimensional perspectives.[Conclusions]The application of computer technologies is of great importance to material research in accelerating progress,improving material understandings,and key technology breakthroughs.

关 键 词：材料信息学 机器学习 数据驱动 

分 类 号：TB30[一般工业技术—材料科学与工程] TP3[自动化与计算机技术—计算机科学与技术]