基于深度学习的晶体性质预测研究

doi:10.16182/j.issn1004731x.joss.21-FZ0772

系统仿真学报 ›› 2021, Vol. 33 ›› Issue (12): 2854-2863.doi: 10.16182/j.issn1004731x.joss.21-FZ0772

基于深度学习的晶体性质预测研究

王步维¹, 王敏², 范谦¹, 王雅男¹, 章涵文¹, 乐云亮^1,*

1.扬州大学信息工程学院,江苏扬州 225127;
2.河北科技大学信息科学与工程学院,河北石家庄 050018

收稿日期:2021-06-13 修回日期:2021-07-29 出版日期:2021-12-18 发布日期:2022-01-13
通讯作者: 乐云亮(1989-),男,博士,讲师,研究方向为机器学习,自旋电子学和电离辐射损伤等。E-mail：yueyunliang@yzu.edu.cn
作者简介:王步维(1997-),男,硕士生,研究方向为磁性半导体和人工智能。E-mail：mz120190680@yzu.edu.cn
基金资助:
江苏省青年基金(BK20190878); 江苏省高等学校自然科学研究面上项目(19KJB510062)

Study on Prediction of Crystal Properties Based on Deep Learning

Wang Buwei¹, Wang Min², Fan Qian¹, Wang Ya'nan¹, zhang hanwen¹, Yue Yunliang^1,*

1. College of Information Engineering, Yangzhou University, Yangzhou 225127, China;
2. School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018, China

Received:2021-06-13 Revised:2021-07-29 Online:2021-12-18 Published:2022-01-13

摘要/Abstract

摘要： 使用传统的机器学习方法预测晶体性质需要进行复杂的特征工程。为了绕过耗时的特征工程,使用基于深度学习技术的元素网络、化学计量比的表征学习、基于注意力的成分限制网络和晶体图卷积神经网络对晶体的形成能、平均每个原子的总能量、带隙和费米能进行模拟。将残差学习引入晶体图卷积神经网络中,提出了一种晶体图卷积残差神经网络。在网络中增加隐藏层的层数和隐藏层的节点个数,通过残差连接的方式连接各个隐藏层,同时加入BatchNorm层进行归一化。经测试发现,相比于晶体图卷积神经网络,晶体图卷积残差神经网络对4种物理量预测的精度提升了1.3%~4.8%,有利于对理想晶体材料进行快速和准确的预测。

关键词: 深度学习, 晶体图卷积, 残差学习, 晶体性质

Abstract: Predicting crystal properties using traditional machine learning methods requires complex feature engineering. In order to bypass time-consuming feature engineering, element network (ElemNet), representation learning from stoichiometry (Roost), compositionally-restricted attention-based network (CrabNet) and crystal graph convolution neural network (CGCNN) based on deep learning technology are used to simulate the formation energy, total energy per atom, band gap, and Fermi energy of crystal. The residual learning is introduced into CGCNN, and a crystal graph convolution residual neural network (CGCRN) is proposed. In the CGCRN, the number of hidden layers and the number of nodes in the hidden layers are increased, and the hidden layers are connected by residual connection. Meanwhile, the BatchNorm layers are added for normalization. Compared with CGCNN, the prediction accuracy of 4 physical quantities using CGCRN has increased by 1.3%~4.8%, which is conducive to rapid and accurate prediction of ideal crystal materials.

Key words: deep learning, crystal graph convolution, residual learning, crystal properties

中图分类号:

TP183
TP391

王步维, 王敏, 范谦, 王雅男, 章涵文, 乐云亮. 基于深度学习的晶体性质预测研究[J]. 系统仿真学报, 2021, 33(12): 2854-2863.

Wang Buwei, Wang Min, Fan Qian, Wang Ya'nan, zhang hanwen, Yue Yunliang. Study on Prediction of Crystal Properties Based on Deep Learning[J]. Journal of System Simulation, 2021, 33(12): 2854-2863.

参考文献

[1] Yue Y, Wang B, Miao N, et al.Tuning the Magnetic Properties of Zr₂N MXene by Biaxial Strain[J]. Ceramics International (S0272-8842), 2021, 47(2): 2367-2373.
[2] Si C, Zhou J, Sun Z.Half-Metallic Ferromagnetism and Surface Functionalization-Induced Metal-Insulator Transition in Graphene-Like Two-Dimensional Cr₂C Crystals[J]. ACS Applied Materials & Interfaces (S1944-8244), 2015, 7(31): 17510-17515.
[3] He M, Zhang L.Machine Learning and Symbolic Regression Investigation on Stability of MXene Materials[J]. Computational Materials Science (S0927-0256), 2021, 196: 110578.
[4] Lu S, Zhou Q, Ouyang Y, et al.Accelerated Discovery of Stable Lead-Free Hybrid Organic-Inorganic Perovskites Via Machine Learning[J]. Nature Communications (S2041-1723), 2018, 9(1): 3405.
[5] Tao Q, Lu T, Sheng Y, et al.Machine Learning Aided Design of Perovskite Oxide Materials for Photocatalytic Water Splitting[J]. Journal of Energy Chemistry (S2095-4956), 2021, 60: 351-359.
[6] Xue D, Balachandran P V, Hogden J, et al.Accelerated Search for Materials with Targeted Properties by Adaptive Design[J]. Nature Communications (S2041-1723), 2016, 7(1): 11241.
[7] Isayev O, Oses C, Toher C, et al.Universal Fragment Descriptors for Predicting Properties of Inorganic Crystals[J]. Nature Communications (S2041-1723), 2017, 8(1): 15679.
[8] Seko A, Togo A, Hayashi H, et al.Prediction of Low-Thermal-Conductivity Compounds with First-Principles Anharmonic Lattice-Dynamics Calculations and Bayesian Optimization[J]. Physical Review Letters (S0031-9007), 2015, 115(20): 205901.
[9] Jha D, Ward L, Paul A, et al.ElemNet: Deep Learning the Chemistry of Materials from Only Elemental Composition[J]. Scientific Reports (S2045-2322), 2018, 8(1): 17593.
[10] Goodall R E A, Lee A A. Predicting Materials Properties Without Crystal Structure: Deep Representation Learning from Stoichiometry[J]. Nature Communications (S2041-1723), 2020, 11(1): 6280.
[11] Wang A Y-T a K, Steven K, Murdock Ryan J, et al. Compositionally Restricted Attention-based Network for Materials Property Predictions[J]. Npj Computational Materials (S2057-3960), 2020, 7(1): 77.
[12] Xie T, Grossman J C.Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties[J]. Physical Review Letters (S0031-9007), 2018, 120(14): 145301.
[13] Kaiming He X Z, Shaoqing Ren, Jian Sun. Deep Residual Learning for Image Recognition[C]// IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, N.J.: IEEE, 2016: 770-778.
[14] Jain A, Ong S P, Hautier G, et al.A Commentary: The Materials Project: A Materials Genome Approach to Accelerating Materials Innovation[J]. APL Materials (S2166-532X), 2013, 1(1): 011002.
[15] Ward L, Agrawal A, Choudhary A, et al.A General-Purpose Machine Learning Framework for Predicting Properties of Inorganic Materials[J]. Npj Computational Materials (S2057-3960), 2016, 2(1): 16028.
[16] Gilmer J, Schoenholz S S, Riley P F, et al.Neural Message Passing for Quantum Chemistry[C]// 34th International Conference on Machine Learning. New York: ACM, 2017: 1263-1272.
[17] Tshitoyan V, Dagdelen J, Weston L, et al.Unsupervised Word Embeddings Capture Latent Knowledge from Materials Science Literature[J]. Nature (S0028-0836), 2019, 571(7763): 95-98.
[18] Tan Z, Wang M, Xie J, et al.Deep Semantic Role Labeling with Self-Attention[C]// 32th AAAI Conference on Artificial Intelligence. Menlo Park, CA: AAAI, 2018: 4929-4936.
[19] Marcheggiani D, Titov I.Encoding Sentences with Graph Convolutional Networks for Semantic Role Labeling[C]// 2017 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2017: 1506-1515.
[20] Szegedy I, Christian S.Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift[C]// 32nd International Conference on Machine Learning. New York: ACM, 2015: 448-456.
[21] Bottou L, Curtis F E, Nocedal J.Optimization Methods for Large-Scale Machine Learning[J]. SIAM Review (S0036-1445), 2018, 60(2): 223-311.
[22] Saal J E, Kirklin S, Aykol M, et al.Materials Design and Discovery with High-Throughput Density Functional Theory: The Open Quantum Materials Database (OQMD)[J]. JOM (S1047-4838), 2013, 65(11): 1501-1509.
[23] Kirklin S, Saal J E, Meredig B, et al.The Open Quantum Materials Database (OQMD): Assessing the Accuracy of DFT Formation Energies[J]. npj Computational Materials (S2057-3960), 2015, 1(1): 864-871.
[24] Yue Y, Zhu H, Liu X, et al.First-Principles Study on Non-Radiative Carrier Captures of Point Defects Associated with Proton Generation in Silica[J]. AIP Advances (S2158-3226), 2021, 11(1): 15214.
[25] Yue Y, Li P, Song Y, et al.Dissociation Characteristics of Proton Release in a-SiO₂ by First-Principles Theory[J]. Journal of Non-Crystalline Solids (S0022-3093), 2018, 486: 1-8.

基于深度学习的晶体性质预测研究

Study on Prediction of Crystal Properties Based on Deep Learning

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	康旭, 张晓峰. 基于生成对抗神经网络的雷达遥感数据增广方法[J]. 系统仿真学报, 2022, 34(4): 920-927.
[2]	林硕, 安磊, 高治军, 单丹, 尚文利. 结合栈式自编码及长短时记忆的入侵检测研究[J]. 系统仿真学报, 2021, 33(6): 1288-1296.
[3]	程文聪, 史小康, 王志刚. 基于生成对抗网络的仿真卫星云图生成方法[J]. 系统仿真学报, 2021, 33(6): 1297-1306.
[4]	董书琴, 张斌. 面向不平衡数据的网络流量异常检测方法[J]. 系统仿真学报, 2021, 33(3): 679-689.
[5]	王霄汉, 张霖, 任磊, 谢堃钰, 王昆玉, 叶飞, 陈真. 基于强化学习的车间调度问题研究简述[J]. 系统仿真学报, 2021, 33(12): 2782-2791.
[6]	冯晓, 张辉, 周蕊, 乔璐, 魏东, 李丹丹, 张玉尧, 郑国清. 基于深度学习和籽粒双面特征的玉米品种识别[J]. 系统仿真学报, 2021, 33(12): 2983-2991.
[7]	杜金莲, 李淑飞, 金雪云. 三维烟雾流场超分辨率数据生成网络模型的研究[J]. 系统仿真学报, 2021, 33(10): 2381-2389.
[8]	刘瑞军, 王向上, 张晨, 章博华. 基于深度学习的视觉SLAM综述[J]. 系统仿真学报, 2020, 32(7): 1244-1256.
[9]	阴敬方, 朱登明, 石敏, 王兆其. 基于引导对抗网络的人体深度图像修补方法[J]. 系统仿真学报, 2020, 32(7): 1312-1321.
[10]	戢晓峰, 戈艺澄. 基于深度学习的节假日高速公路交通流预测方法[J]. 系统仿真学报, 2020, 32(6): 1164-1171.
[11]	孔锐, 谢玮, 雷泰. 基于神经网络的图像描述方法研究[J]. 系统仿真学报, 2020, 32(4): 601-611.
[12]	秦胜伟, 李重, 李金锋, 陈梓浩, 丁靖骞, 刘万顺. 校园漫游互动AR系统设计与实现[J]. 系统仿真学报, 2019, 31(7): 1367-1376.
[13]	叶继华, 时淑霞, 李汉曦, 王仕民, 杨思渝. 基于深度学习的驾驶关注区域检测研究与实现[J]. 系统仿真学报, 2019, 31(7): 1421-1428.
[14]	叶继华, 时淑霞, 李汉曦, 左家莉, 王仕民. 一种结合深度学习和随机森林的地平线检测方法[J]. 系统仿真学报, 2018, 30(7): 2507-2514.
[15]	石祥滨, 钟健, 刘翠微, 刘芳, 张德园. 小样本高光谱遥感图像深度学习方法[J]. 系统仿真学报, 2018, 30(7): 2744-2752.