深度学习：材料微结构与物性研究中的新动力
Deep Learning: New Engine for the Study of Material Microstructures and Physical Properties
DOI: 10.12677/MP.2019.96026, PDF,  被引量    科研立项经费支持
作者: 卢 果, 段素青：北京应用物理与计算数学研究所，计算物理重点实验室，北京
关键词: 深度学习；材料微结构与物性；多尺度模拟；Deep Learning； Material Microstructures and Materials Properties； Multiscale Simulation
摘要: 材料的微结构决定了其宏观性质。传统的自下而上的多尺度方法提供了研究微结构和物性关联的整体思路，但针对材料的微观、介观和宏观建模都存在诸多困难，其中不同尺度描述的衔接就极具挑战。随着计算能力以摩尔定律式的提升，和人工智能特别是深度学习的爆炸式进展，基于数据驱动的材料研究成当今世界的研究热点，并在晶体结构预测、稳定性分析、状态方程、光学性质、化学合成等领域都取得了很好的应用效果。深度学习的快速计算和可靠的预测能力能够极大的提升材料模拟效率。其广泛适用性为材料微结构与多尺度模拟中的一些传统难题提供了新的研究思路，有望促进材料微结构与物性的研究，为基于微介观机理的宏观物性建模、满足工程应用要求的材料性质预测提供了新的研究方向。本文将简要介绍深度学习的原理及常用的神经网络类型，概述材料微结构与多尺度建模中的主要方法，随后介绍深度学习在材料微结构与物性研究中的进展，并展望该方法在材料多尺度模拟领域中的发展。
Abstract: The microstructures of materials determine their macroscopic properties. The traditional bottom-up multi-scale approach provides a general strategy for studying the relationship between microstructures and physical properties. However, there are still many difficulties in microscopic, mesoscopic and macroscopic modeling of materials, and the bridging of different scale models is extremely challenging. With the advancement of computing power in Moore’s Law, and the explosive development of artificial intelligence, especially deep learning, data-driven methods are commonly used, and in crystal structure prediction, stability analysis, equation of states, optical properties, chemical synthesis, etc. have achieved good application results. The fast calculation speed and reliable prediction capabilities of deep learning can greatly improve the efficiency of material simulation. Its wide applicability provides new research ideas for some traditional problems in material microstructures and multi-scale simulations. It is expected to promote the study of material microstructures and physical properties, and to provide new research directions for modelling macroscopic physical properties based on micro-mesoscopic mechanism and prediction of material properties to meet engineering application requirements. This review article will briefly introduce the basic principles of deep learning and main types of commonly used neural networks, outline the main methods of material microstructures and multi-scale modeling, and then introduce the recent progress of deep learning method in the study of material microstructure and physical properties, and review the developments and prospects of deep learning method in the field of multi-scale simulation of materials.
文章引用：卢果, 段素青. 深度学习：材料微结构与物性研究中的新动力[J]. 现代物理, 2019, 9(6): 263-276. https://doi.org/10.12677/MP.2019.96026

参考文献

[1] McDowell, D.L. and LeSar, R.A. (2016) The Need for Microstructure Informatics in Process-Structure-Property Relations. MRS Bulletin, 41, 587-593.
[Google Scholar] [CrossRef] 
[2] Ashby, M.F. (1992) Physical Modelling of Materials Problems. Materials Science and Technology, 8, 102-111.
[Google Scholar] [CrossRef] 
[3] Raabe, D. (1998) Computational Materials Science: The Simulation of Materials Microstructures and Properties. Wiley-VCH, Hoboken.
[Google Scholar] [CrossRef] 
[4] McDowell, D.L. (2018) Multiscale Crystalline Plasticity for Materials Design. In: Shin, D. and Saal, J., Eds., Computational Materials System Design, Springer International Publishing, Berlin, 105-146.
[Google Scholar] [CrossRef] 
[5] McDowell, D.L. (2019) Multiscale Modeling of Interfaces, Dislocations, and Dislocation Field Plasticity. In: Mesarovic, S., Forest, S. and Zbib, H., Eds., Mesoscale Models: From Micro-Physics to Macro-Interpretation, Springer International Publishing, Berlin, 195-297.
[Google Scholar] [CrossRef] 
[6] Butler, K.T., Davies, D.W., Cartwright, H., Isayev, O. and Walsh, A. (2018) Machine Learning for Molecular and Materials Science. Nature, 559, 547.
[Google Scholar] [CrossRef] [PubMed]
[7] Ramprasad, R., Batra, R., Pilania, G., Mannodi-Kanakkithodi, A. and Kim, C. (2017) Machine Learning in Materials Informatics: Recent Applications and Prospects. NPJ Computational Materials, 3, 54.
[Google Scholar] [CrossRef] 
[8] Goh, G.B., Hodas, N.O. and Vishnu, A. (2017) Deep Learning for Computational Chemistry. Journal of Computational Chemistry, 38, 1291-1307.
[Google Scholar] [CrossRef] [PubMed]
[9] Dechter, R. (1986) Learning While Searching in Constraint-Satisfaction-Problems. 5th AAAI, Philadelphia, 11-15 August 1986, 178-185.
[10] Aizenberg, I.N., Aizenberg, N.N. and Vandewalle, J. (2000) Multi-Valued and Universal Binary Neurons: Theory, Learning and Applications. Springer US, Berlin.
[Google Scholar] [CrossRef] 
[11] Bengio, Y. (2009) Learning Deep Architectures for AI. Foundations and Trends® in Machine Learning, 2, 1-127.
[Google Scholar] [CrossRef] 
[12] Schmidhuber, J. (2015) Deep Learning in Neural Networks: An Overview. Neural Networks, 61, 85-117.
[Google Scholar] [CrossRef] [PubMed]
[13] Ryan, K., Lengyel, J. and Shatruk, M. (2018) Crystal Structure Prediction via Deep Learning. Journal of the American Chemical Society, 140, 10158-10168.
[Google Scholar] [CrossRef] [PubMed]
[14] Cassar, D.R., de Carvalho, A.C. and Zanotto, E.D. (2018) Predicting Glass Transition Temperatures Using Neural Networks. Acta Materialia, 159, 249-256.
[Google Scholar] [CrossRef] 
[15] Jha, D., et al. (2018) Element: Deep Learning the Chemistry of Materials from Only Elemental Composition. Scientific Reports, 8, Article No. 17593.
[Google Scholar] [CrossRef] [PubMed]
[16] Yamamoto, T. (2019) Crystal Graph Neural Networks for Data Mining in Materials Science.
[17] Fukushima, K. (2007) Neocognitron. Scholarpedia, 2, 1717.
[Google Scholar] [CrossRef] 
[18] Matsugu, M., Mori, K., Mitari, Y. and Kaneda, Y. (2003) Subject Independent Facial Expression Recognition with Robust Face Detection Using a Convolutional Neural Network. Neural Networks, 16, 555-559.
[Google Scholar] [CrossRef] 
[19] Zhang, W., Itoh, K., Tanida, J. and Ichioka, Y. (1990) Parallel Distributed Processing Model with Local Space-Invariant Interconnections and Its Optical Architecture. Applied Optics, 29, 4790-4797.
[Google Scholar] [CrossRef] 
[20] Zheng, X., Zheng, P. and Zhang, R.-Z. (2018) Machine Learning Material Properties from the Periodic Table Using Convolutional Neural Networks. Chemical Science, 9, 8426-8432.
[Google Scholar] [CrossRef] 
[21] Xie, T. and Grossman, J.C. (2018) Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties. Physical Review Letters, 120, Article ID: 145301.
[Google Scholar] [CrossRef] 
[22] Sanyal, S., et al. (2018) MT-CGCNN: Integrating Crystal Graph Convolutional Neural Network with Multitask Learning for Material Property Prediction.
[23] Rumelhart, D.E., Hinton, G.E. and Williams, R.J. (1986) Learning Representations by Back-Propagating Errors. Nature, 323, 533-536.
[Google Scholar] [CrossRef] 
[24] Hochreiter, S. and Schmidhuber, J. (1997) Long Short-Term Memory. Neural Computation, 9, 1735-1780.
[Google Scholar] [CrossRef] [PubMed]
[25] Quan, Z., et al. (2018) A System for Learning Atoms Based on Long Short-Term Memory Recurrent Neural Networks. IEEE International Conference on Bioinformatics and Biomedicine, Madrid, 3-6 December 2018, 728-733.
[Google Scholar] [CrossRef] 
[26] Goodfellow, I., et al. (2014) Generative Adversarial Nets. Advances in Neural Information Processing Systems, Montreal, 8-13 December 2014, 2672-2680.
[27] Zhao, J., Mathieu, M. and LeCun, Y. (2016) Energy-Based Generative Adversarial Network.
[28] Liu, Y.-H. and van Nieuwenburg, E.P. (2018) Discriminative Cooperative Networks for Detecting Phase Transitions. Physical Review Letters, 120, Article ID: 176401.
[Google Scholar] [CrossRef] 
[29] Hinton, G.E., Osindero, S. and Teh, Y.-W. (2006) A Fast Learning Algorithm for Deep Belief Nets. Neural Computation, 18, 1527-1554.
[Google Scholar] [CrossRef] [PubMed]
[30] Bu, H., Gan, Y., Wang, Y., Zhou, S. and Guan, J. (2006) A New Method for Enhancer Prediction Based on Deep Belief Network. BMC Bioinformatics, 18, Article No. 418.
[Google Scholar] [CrossRef] [PubMed]
[31] Bock, F.E., et al. (2019) A Review of the Application of Machine Learning and Data Mining Approaches in Continuum Materials Mechanics. Frontiers in Materials, 6, 110.
[Google Scholar] [CrossRef] 
[32] Ye, W., Chen, C., Wang, Z., Chu, I.-H. and Ong, S.P. (2018) Deep Neural Networks for Accurate Predictions of Crystal Stability. Nature Communications, 9, Article No. 3800.
[Google Scholar] [CrossRef] [PubMed]
[33] Schmidt, J., et al. (2017) Predicting the Thermodynamic Stability of Solids Combining Density Functional Theory and Machine Learning. Chemistry of Materials, 29, 5090-5103.
[Google Scholar] [CrossRef] 
[34] Desgranges, C. and Delhommelle, J. (2018) A New Approach for the Prediction of Partition Functions Using Machine Learning Techniques. The Journal of Chemical Physics, 149, Article ID: 044118.
[Google Scholar] [CrossRef] [PubMed]
[35] Yassar, R.S., AbuOmar, O., Hansen, E. and Horstemeyer, M.F. (2010) On Dislocation-Based Artificial Neural Network Modeling of Flow Stress. Materials & Design, 31, 3683-3689.
[Google Scholar] [CrossRef] 
[36] Jones, R.E., Templeton, J.A., Sanders, C.M. and Ostien, J.T. (2018) Machine Learning Models of Plastic Flow Based on Representation Theory. Computer Modeling in Engineering & Sciences, 117, 309-342.
[Google Scholar] [CrossRef] 
[37] Stein, H.S., Guevarra, D., Newhouse, P.F., Soedarmadji, E. and Gregoire, J.M. (2019) Machine Learning of Optical Properties of Materials-Predicting Spectra from Images and Images from Spectra. Chemical Science, 10, 47-55.
[Google Scholar] [CrossRef] 
[38] Meredig, B., et al. (2018) Can Machine Learning Identify the Next High-Temperature Superconductor? Examining Extrapolation Performance for Materials Discovery. Molecular Systems Design & Engineering, 3, 819-825.
[Google Scholar] [CrossRef] 
[39] Jančíková, Z., Zimný, O. and Koštial, P. (2013) Prediction of Metal Corrosion by Neural Networks. Metalurgija-Sisak then Zagreb-, 52, 379-381.
[40] Kusiak, J. and Kuziak, R. (2002) Modelling of Microstructure and Mechanical Properties of Steel Using the Artificial Neural Network. Journal of Materials Processing Technology, 127, 115-121.
[Google Scholar] [CrossRef] 
[41] Li, X., et al. (2018) A Transfer Learning Approach for Microstructure Reconstruction and Structure-Property Predictions. Scientific Reports, 8, Article No. 13461.
[Google Scholar] [CrossRef] [PubMed]
[42] Li, X., et al. (2019) Predicting the Effective Mechanical Property of Heterogeneous Materials by Image Based Modeling and Deep Learning. Computer Methods in Applied Mechanics and Engineering, 347, 735-753.
[Google Scholar] [CrossRef] 
[43] Wang, Z.-L. and Adachi, Y. (2019) Property Prediction and Properties-to-Microstructure Inverse Analysis of Steels by a Machine-Learning Approach. Materials Science and Engineering: A, 744, 661-670.
[Google Scholar] [CrossRef] 
[44] Barton, N.R., et al. (2011) A Multiscale Strength Model for Extreme Loading Conditions. Journal of Applied Physics, 109, Article ID: 073501.
[Google Scholar] [CrossRef] 
[45] Barton, N.R. and Rhee, M. (2013) A Multiscale Strength Model for Tantalum over an Extended Range of Strain Rates. Journal of Applied Physics, 114, Article ID: 123507.
[Google Scholar] [CrossRef] 
[46] Cecen, A., Dai, H., Yabansu, Y.C., Kalidindi, S.R. and Song, L. (2018) Material Structure-Property Linkages Using Three-Dimensional Convolutional Neural Networks. Acta Materialia, 146, 76-84.
[Google Scholar] [CrossRef] 
[47] Salmenjoki, H., Alava, M.J. and Laurson, L. (2018) Machine Learning Plastic Deformation of Crystals. Nature Communications, 9, Article No. 5307.
[Google Scholar] [CrossRef] [PubMed]
[48] Kusche, C., et al. (2019) Large-Area, High-Resolution Characterisation and Classification of Damage Mechanisms in Dual-Phase Steel Using Deep Learning. PLoS ONE, 14, e0216493.
[Google Scholar] [CrossRef] [PubMed]
[49] Dengiz, O., Smith, A. and Nettleship, I. (2005) Grain Boundary Detection in Microstructure Images Using Computational Intelligence. Computers in Industry, 56, 854-866.
[Google Scholar] [CrossRef] 
[50] Albuquerque, V.H.C., Cortez, P.C., Alexandria, A.R. and Tavares, J.M.R.S. (2008) A New Solution for Automatic Microstructures Analysis from Images Based on a Backpropagation Artificial Neural Network. Nondestructive Testing and Evaluation, 23, 273-283.
[Google Scholar] [CrossRef] 
[51] Paulic, M., et al. (2015) Intelligent System for Prediction of Mechanical Properties of Material Based on Metallographic Images. Tehničkivjesnik, 22, 1419-1424.
[Google Scholar] [CrossRef] 
[52] Mulewicz, B., Korpala, G., Kusiak, J. and Prahl, U. (2019) Autonomous Interpretation of the Microstructure of Steels and Special Alloys. Materials Science Forum, 949, 24-31.
[Google Scholar] [CrossRef] 
[53] Adachi, Y., Taguchi, M. and Hirokawa, S. (2016) Microstructure Recognition by Deep Learning. Tetsu-to-Hagane, 102, 722-729.
[Google Scholar] [CrossRef] 
[54] Ruelas-Santoyo, E.A., et al. (2018) System for the Recognition of Wear Patterns on Microstructures of Carbon Steels Using a Multilayer Perceptron. Ingeniería e Investigación, 38, 113-120.
[Google Scholar] [CrossRef] 
[55] Azimi, S.M., Britz, D., Engstler, M., Fritz, M. and Mücklich, F. (2018) Advanced Steel Microstructural Classification by Deep Learning Methods. Scientific Reports, 8, Article No. 2128.
[Google Scholar] [CrossRef] [PubMed]
[56] Wang, B., Yager, K., Yu, D. and Hoai, M. (2017) X-Ray Scattering Image Classification Using Deep Learning. IEEE Winter Conference on Applications of Computer Vision, Santa Rosa, 24-31 March 2017, 697-704.
[Google Scholar] [CrossRef] 
[57] DeCost, B.L., Lei, B., Francis, T. and Holm, E.A. (2019) High Throughput Quantitative Metallography for Complex Microstructures Using Deep Learning: A Case Study in Ultrahigh Carbon Steel. Microscopy and Microanalysis, 25, 21-29.
[Google Scholar] [CrossRef] 
[58] Li, S., Sinha, A., Lee, J. and Barbastathis, G. (2018) Quantitative Phase Microscopy Using Deep Neural Networks. Quantitative Phase Imaging IV, San Francisco, 27 January-1 February 2018, 84.
[Google Scholar] [CrossRef] 
[59] Li, W., Field, K.G. and Morgan, D. (2018) Automated Defect Analysis in Electron Microscopic Images. NPJ Computational Materials, 4, 36.
[Google Scholar] [CrossRef] 
[60] Maksov, A., et al. (2019) Deep Learning Analysis of Defect and Phase Evolution during Electron Beam-Induced Transformations in WS2. NPJ Computational Materials, 5, 12.
[Google Scholar] [CrossRef] 
[61] Behler, J. and Parrinello, M. (2007) Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces. Physical Review Letters, 98, Article ID: 146401.
[Google Scholar] [CrossRef] 
[62] Zhang, L., Han, J., Wang, H., et al. (2018) Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics. Physical Review Letters, 120, Article ID: 143001.
[Google Scholar] [CrossRef] 
[63] Wang, H., Zhang, L., Han, J. and E, W. (2018) DeePMD-Kit: A Deep Learning Package for Many-Body Potential Energy Representation and Molecular Dynamics. Computer Physics Communications, 228, 178-184.
[Google Scholar] [CrossRef] 
[64] Zhang, L., Lin, D.-Y., Wang, H., Car, R. and E, W. (2019) Active Learning of Uniformly Accurate Inter-Atomic Potentials for Materials Simulation. Physical Review Materials, 3, Article ID: 023804.
[Google Scholar] [CrossRef] 
[65] Chan, H., et al. (2019) Machine Learning Classical Interatomic Potentials for Molecular Dynamics from First-Principles Training Data. The Journal of Physical Chemistry C, 123, 6941-6957.
[Google Scholar] [CrossRef] 
[66] Endo, K. (2018) Multi-Step Time Series Generator for Molecular Dynamics. 32nd AAAI Conference on Artificial Intelligence, 2-7 February 2018.
[67] Xie, T., France-Lanord, A., Wang, Y., Shao-Horn, Y. and Grossman, J.C. (2019) Graph Dynamical Networks for Unsupervised Learning of Atomic Scale Dynamics in Materials. Nature Communications, 10, Article No. 2667.
[Google Scholar] [CrossRef] [PubMed]
[68] Salmenjoki, H. (2017) Predicting the Behaviour of Dislocation Systems with Machine Learning Methods.
[69] Salmenjoki, H., Alava, M. and Laurson, L. (2018) Mimicking Complex Dislocation Dynamics by Interaction Networks. The European Physical Journal B, 91, 275.
[Google Scholar] [CrossRef] 
[70] Zhang, Y. and Ngan, A.H.W. (2019) Extracting Dislocation Microstructures by Deep Learning. International Journal of Plasticity, 115, 18-28.
[Google Scholar] [CrossRef] 
[71] Nomoto, S., Segawa, M. and Wakameda, H. (2018) Non-Equilibrium Phase Field Model Using Thermodynamics Data Estimated by Machine Learning for Additive Manufacturing Solidification. Solid Freeform Fabrication Proceedings, Austin.
[72] Goswami, S., Anitescu, C., Chakraborty, S. and Rabczuk, T. (2019) Transfer Learning Enhanced Physics Informed Neural Network for Phase-Field Modeling of Fracture.
[73] Mao, X., Joshi, V., Miyanawala, T.P. and Jaiman, R.K. (2018) Data-Driven Computing with Convolutional Neural Networks for Two-Phase Flows: Application to Wave-Structure Interaction. Proceedings of the ASME 2018 37th International Conference on Ocean, Offshore and Arctic Engineering, Volume 2, Madrid, 17-22 June 2018, V002T08A005.
[Google Scholar] [CrossRef] 
[74] Gibou, F., Hyde, D. and Fedkiw, R. (2019) Sharp Interface Approaches and Deep Learning Techniques for Multiphase Flows. Journal of Computational Physics, 380, 442-463.
[Google Scholar] [CrossRef]