介科学引导的深度学习案例研究

郭力; 孟凡勇; 秦鹏飞; 夏诏杰; 常麒; 陈建华; 李静海

doi:10.1016/j.eng.2024.01.007

PDF(1745 KB)

工程（英文） ›› 2024, Vol. 39 ›› Issue (8) : 84-93. DOI: 10.1016/j.eng.2024.01.007

研究论文

Article

介科学引导的深度学习案例研究

郭力 ^a^,^b^,^* ,
孟凡勇 ^a ,
秦鹏飞 ^a^,^b ,
夏诏杰 ^a^,^b ,
常麒 ^a ,
陈建华 ^a ,
李静海 ^a

作者信息 +

A Case Study Applying Mesoscience to Deep Learning

Li Guo ^a^,^b^,^* ,
Fanyong Meng ^a ,
Pengfei Qin ^a^,^b ,
Zhaojie Xia ^a^,^b ,
Qi Chang ^a ,
Jianhua Chen ^a ,
Jinghai Li ^a

Author information +

History +

Abstract

In this paper, we propose mesoscience-guided deep learning (MGDL), a deep learning modeling approach guided by mesoscience, to study complex systems. When establishing sample dataset based on the same system evolution data, different from the operation of conventional deep learning method, MGDL introduces the treatment of the dominant mechanisms of complex system and interactions between them according to the principle of compromise in competition (CIC) in mesoscience. Mesoscience constraints are then integrated into the loss function to guide the deep learning training. Two methods are proposed for the addition of mesoscience constraints. The physical interpretability of the model-training process is improved by MGDL because guidance and constraints based on physical principles are provided. MGDL was evaluated using a bubbling bed modeling case and compared with traditional techniques. With a much smaller training dataset, the results indicate that mesoscience-constraint-based model training has distinct advantages in terms of convergence stability and prediction accuracy, and it can be widely applied to various neural network configurations. The MGDL approach proposed in this paper is a novel method for utilizing the physical background information during deep learning model training. Further exploration of MGDL will be continued in the future.

Keywords

Mesoscience / Deep learning / Complex system / Gas-solid system / Bubbling bed

引用本文

EndNote

Ris (Procite)

Bibtex

导出引用

郭力, 孟凡勇, 秦鹏飞. 将介科学应用于深度学习的案例研究. Engineering. 2024, 39(8): 84-93 https://doi.org/10.1016/j.eng.2024.01.007

参考文献

原文顺序 | 文献年度倒序 | 文中引用次数倒序

[1]	G.E. Karniadakis, I.G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, L. Yang. Physics-informed machine learning. Nat Rev Phys, 3 (6) (2021), pp. 422-440.
[2]	M. Raissi, P. Perdikaris, G.E. Karniadakis. Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. J Comput Phys, 378 (2019), pp. 686-707.
[3]	S.J. Raymond, N.J. Cecchi, H.V. Alizadeh, A.A. Callan, E. Rice, Y. Liu, et al. Physics-informed machine learning improves detection of head impacts. Ann Biomed Eng, 50 (11) (2022), pp. 1534-1545.
[4]	X. Jiang, D. Wang, Q. Fan, M. Zhang, C. Lu, A.P.T. Lau. Physics-informed neural network for nonlinear dynamics in fiber optics. Laser Photonics Rev, 16 (9) (2022), p. 2100483.
[5]	S. Sharma, R. Awasthi, Y.S. Sastry, P.R. Budarapu. Physics-informed neural networks for estimating stress transfer mechanics in single lap joints. J Zhejiang Univ Sci A, 22 (8) (2021), pp. 621-631.
[6]	Z.P. Mao, A.D. Jagtap, G.E. Karniadakis. Physics-informed neural networks for high-speed flows. Comput Methods Appl Mech Eng, 360 (2020), Article 112789.
[7]	S. Chakraborty. Transfer learning based multi-fidelity physics informed deep neural network. J Comput Phys, 426 (2021), Article 109942.
[8]	X. Liu, X. Zhang, W. Peng, W. Zhou, W. Yao. A novel meta-learning initialization method for physics-informed neural networks. Neural Comput Appl, 34 (17) (2022), pp. 14511-14534.
[9]	J.H. Li, W.L. Huang. Towards mesoscience: the principle of compromise in competition. Springer, Heidelberg (2014).
[10]	J.H. Li, W.L. Huang, J.H. Chen, W. Ge, C.F. Hou. Mesoscience based on the EMMS principle of compromise in competition. Chem Eng J, 333 (2018), pp. 327-335.
[11]	J.H. Li. Exploring the logic and landscape of the knowledge system: multilevel structures, each multiscaled with complexity at the mesoscale. Engineering, 2 (3) (2016), pp. 276-285.
[12]	L. Guo, J. Wu, J.H. Li. Complexity at mesoscales: a common challenge in developing artificial intelligence. Engineering, 5 (5) (2019), pp. 924-929.
[13]	J.H. Li, Y.K. Tung, M.S. Kwauk. Method of energy minimization in multi-scale modeling of particle-fluid two-phase flow. P. Basu, J.F. Large (Eds.), Circulating fluidized bed technology II, Pergamon Press, New York (1988), pp. 89-103.
[14]	J.H. Li, J.Y. Zhang, W. Ge, X.H. Liu. Multi-scale methodology for complex systems. Chem Eng Sci, 59 (8-9) (2004), pp. 1687-1700.
[15]	L.M. Wang, X.P. Qiu, L. Zhang, J.H. Li. Turbulence originating from the compromise-in-competition between viscosity and inertia. Chem Eng J, 300 (2016), pp. 83-97.
[16]	W.L. Huang, J.H. Li. Mesoscale model for heterogeneous catalysis based on the principle of compromise in competition. Chem Eng Sci, 147 (2016), pp. 83-90.
[17]	J.H. Chen, Y. Ren, W.L. Huang, L. Zhang, J.H. Li. Multilevel mesoscale complexities in mesoregimes: challenges in chemical and biochemical engineering. Annu Rev Chem Biomol Eng, 13 (1) (2022), pp. 431-455.
[18]	J.Y. Zhang, W. Ge, J.H. Li. Simulation of heterogeneous structures and analysis of energy consumption in particle-fluid systems with pseudo-particle modeling. Chem Eng Sci, 60 (11) (2005), pp. 3091-3099.
[19]	L. Zhang, J.H. Chen, W.L. Huang, J.H. Li. A direct solution to multi-objective optimization: validation in solving the EMMS model for gas-solid fluidization. Chem Eng Sci, 192 (2018), pp. 499-506.
[20]	H.B. Mann. Nonparametric tests against trend. Econometrica, 13 (3) (1945), pp. 245-259.
[21]	M.G. Kendall. Rank correlation methods. (4th ed.), Charless Griffin, London (1975).
[22]	D.R. Cox, A. Stuart. Some quick sign tests for trend in location and dispersion. Biometrika, 42 (1-2) (1955), pp. 80-95.
[23]	A.M. Squires. The story of fluid catalytic cracking: the first “circulating fluidized beds”. P. Basu (Ed.), Circulating fluidized bed technology, Pergamon Press, New York (1986), pp. 1-19.
[24]	L. Reh. New and efficient high-temperature processes with circulating fluid-bed reactors. Chem Eng Technol, 18 (2) (1995), pp. 75-89.
[25]	L. Lu, X. Gao, J.F. Dietiker, M. Shahnam, W.A. Rogers. Machine learning accelerated discrete element modeling of granular flows. Chem Eng Sci, 245 (2021), Article 116832.
[26]	Z. Yang, B.N. Lu, W. Wang. Coupling artificial neural network with EMMS drag for simulation of dense fluidized beds. Chem Eng Sci, 246 (2021), Article 117003.
[27]	H. Bazai, E. Kargar, M. Mehrabi. Using an encoder-decoder convolutional neural network to predict the solid holdup patterns in a pseudo-2D fluidized bed. Chem Eng Sci, 246 (2021), Article 116886.
[28]	B. Ouyang, L.T. Zhu, Y.H. Su, Z.H. Luo. A hybrid mesoscale closure combining CFD and deep learning for coarse-grid prediction of gas-particle flow dynamics. Chem Eng Sci, 248 (Pt B) ( 2022), Article 117268.
[29]	P.F. Qin, Z.J. Xia, L. Guo. A deep learning approach using temporal-spatial data of computational fluid dynamics for fast property prediction of gas-solid fluidized bed. Korean J Chem Eng, 40 (1) (2023), pp. 57-66.
[30]	M. Upadhyay, V.M. Nagulapati, H. Lim. Hybrid CFD-neural networks technique to predict circulating fluidized bed reactor riser hydrodynamics. J Clean Prod, 337 (2022), Article 130490.
[31]	D. Gidaspow. Multiphase flow and fluidization:continuum and kinetic theory descriptions. Academic Press, New York (1994).
[32]	Y. Tsuji, T. Tanaka, T. Ishida. Lagrangian numerical simulation of plug flow of cohesionless particles in a horizontal pipe. Powder Technol, 71 (3) (1992), pp. 239-250.
[33]	S. Tenneti, S. Subramaniam. Particle-resolved direct numerical simulation for gas-solid flow model development. Annu Rev Fluid Mech, 46 (1) (2014), pp. 199-230.
[34]	H.H. Cui, Q. Chang, J.H. Chen, W. Ge. PR-DNS verification of the stability condition in the EMMS model. Chem Eng J, 401 (2020), Article 125999.
[35]	C.S. Peskin. Numerical analysis of blood flow in the heart. J Comput Phys, 25 (3) (1977), pp. 220-252.
[36]	J. Donahue, L.A. Hendricks, M. Rohrbach, S. Venugopalan, S. Guadarrama, K. Saenko, et al. Long-term recurrent convolutional networks for visual recognition and description. IEEE Trans Pattern Anal Mach Intell, 39 (4) (2017), pp. 677-691.
[37]	Shi XJ, Chen ZR, Wang H, Yeung DY, Wong WK, Woo WC. Convolutional LSTM network: a machine learning approach for precipitation nowcasting. In: Cortes C, Lee DD, Sugiyama M, Garnett R, editors. Proceedings of the 28th International Conference on Neural Information Processing Systems; 2015 Dec 7- 12; Montreal, QC, Canada; 2015. p. 802-10.
[38]	Mathieu MM, Couprie C, LeCun Y. Deep multi-scale video prediction beyond mean square error. 2015. arXiv:1511.05440.
[39]	Z. Wang, A.C. Bovik, H.R. Sheikh, E.P. Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE Trans Image Process, 13 (4) (2004), pp. 600-612.
[40]	Li JH. Insight: the journey ahead for AI [Internet]. Norwich: Business Chief; 2020 May 20 [cited 2024 Jan 16]. Available from: