高分子材料基因组——高分子研发的新方法  

Polymer Genome Approach:A New Method for Research and Development of Polymers

作　　者：都仕 张宋奇 王立权[1] 林嘉平[1] 杜磊[1] Shi Du;Song-qi Zhang;Li-quan Wang;Jia-ping Lin;Lei Du(School of Materials Science and Engineering,East China University of Science and Technology,Shanghai 200237)

机构地区：[1]华东理工大学材料科学与工程学院,上海200237

出　　处：《高分子学报》2022年第6期592-607,共16页Acta Polymerica Sinica

基　　金：国家自然科学基金(基金号51833003,22173030,21975073,21774032,51621002)资助项目.

摘　　要：材料基因组旨在将计算工具、数据库和实验工具有机结合,缩短研发时间,提高材料研发效率.高分子材料因其结构独特性和复杂性阻碍了材料基因组在高分子材料领域的发展.目前,国内外学者在高通量筛选高分子化学结构策略和构建性能预测方法等方面开展了尝试,并取得了一些成果.本文总结和评述了当前利用代理量方法和机器学习预测模型实现高分子材料基因组的进展,利用可计算的量代理宏观性能的代理量法和利用机器学习模型预测材料性能的方法在一定程度上克服了高分子复杂性的影响.在此基础上,系统地介绍了数据挖掘或模型构建的方法以及运用这些模型筛选不同类型高分子的思路,着重探讨了方法构建和材料筛选背后的思想以及对各类问题的解决措施.最后,探讨了当前高分子材料基因组发展中所面临的主要挑战,并展望了高分子材料基因组的未来发展方向.The materials genome approach(MGA),which can accelerate the research and development of new materials via combining computer technology,database technology,and experiments,has attracted considerable attention from academia and industry.However,establishing the structure-property relationship of polymer is more complicated than those of metal and inorganic materials because of the complex structural features such as chain architecture,chain configuration,chain conformation,and chain aggregation.The difficulty in building the structure-property relationship has hindered the development of MGA in polymers.Recently,there have been increasing studies on the rational design of advanced polymers by MGA.This review summarizes research progresses on the MGA of polymer,including establishing structure-property relationships that can predict polymer properties,exploring the vast chemical space of polymers,and rationally designing polymer structures.Calculating the key features that correlate with the desired properties from data mining is one of the ways to screen promising polymers.Alternatively,machine learning can construct structure-property relationships automatically based on databases.The model based on machine learning can apply to the forward and inverse design of advanced polymers.These two prevalent methods are presented.The review systematically introduces the methods of data mining or model construction and the ideas of screening different types of polymers by using these models and focuses on the ideas behind method construction and material screening and the solutions to various problems.In addition,the challenges faced by the development of polymer MGA are also outlined.To push forward the research on polymer MGA,we suggest paying more attention to introducing more efficient machine learning technology,establishing more comprehensive polymer databases,and developing highthroughput experimental technology in the future.

关 键 词：材料基因组 高分子 机器学习 理论模拟 

分 类 号：TQ317[化学工程—高聚物工业] Q789[生物学—分子生物学]