Abstract
Numerical simulation in Computational Fluid Dynamics mainly relies on discretizing the governing equations in time or space to obtain numerical solutions, which is expensive and time-consuming. Deep learning significantly reduces the computational cost due to its great nonlinear curve fitting capability, however, the data-driven models is agnostic to latent relationships between input and output. In this paper, a novel deep learning named Navier–Stokes Generative Adversarial Network integrated with physical information is proposed. The Navier–Stokes Equation is added to the loss function of the generator in the form of residuals, which means physics loss in this paper. Then, the proposed model is trained in the framework of generative adversarial network. Experimental results show that proposed model significantly outperform similar models, mean absolute error are decreased by 62.29, 78.42 and 78.61% on pressure and velocity components. What’s more, effectiveness of introducing physics loss is also verified.
Similar content being viewed by others
References
Bai Xd, Wang Y, Zhang W (2020) Applying physics informed neural network for flow data assimilation. J Hydrodyn. pp. 1–9
Baydin AG, Pearlmutter BA, Radul AA, Siskind JM (2018) Automatic differentiation in machine learning: a survey. J Mach Learn Res. Vol. 18
Booz J, Yu W, Xu G, Griffith D, Golmie N (2019) A deep learning-based weather forecast system for data volume and recency analysis. In: 2019 International Conference on Computing, Networking and Communications (ICNC), pp. 697–701. IEEE
Erichson NB, Muehlebach M, Mahoney MW (2019) Physics-informed autoencoders for lyapunov-stable fluid flow prediction. arXiv preprint arXiv:1905.10866
Han Y, Chun Q, Jin H (2021) Wind-induced vibration performance of early chinese hall-style timber buildings. J Wood Sci 67(1):1–18
Hornik K, Stinchcombe M, White H (1989) Multilayer feedforward networks are universal approximators. Neural Netw 2(5):359–366
Huang G, Jafari AH (2021) Enhanced balancing gan: minority-class image generation. Neural Comput Appl. pp. 1–10
Islama M, Nasrinb S (2020) Dusty fluid flow past between two parallel riga plates embedded in a porous medium
Isola P, Zhu JY, Zhou T, Efros AA (2016) Image-to-image translation with conditional adversarial networks
Kim B, Azevedo VC, Thuerey N, Kim T, Gross M, Solenthaler B (2019) Deep fluids: a generative network for parameterized fluid simulations. In: Computer Graphics Forum, vol. 38, pp. 59–70. Wiley Online Library
Lye KO, Mishra S, Ray D (2020) Deep learning observables in computational fluid dynamics. J Comput Phys 410:109339
Major R, Kopernik M, Kuźmińska A, Imbir G, Plutecka H, Pomorska M, Ciach T, Lackner JM (2021) In vitro haemocompatibility assessment of acrylic acid deposited on solid, polyurethane substrate. Colloids Surf, B 199:111562
Mao S, Rajan D, Chia LT (2021) Deep residual pooling network for texture recognition. Pattern Recogn 112:107817
Noé F, Olsson S, Köhler J, Wu H (2019) Boltzmann generators: sampling equilibrium states of many-body systems with deep learning. Science 365:6457
Obukhov AD, Krasnyanskiy MN (2021) Neural network method for automatic data generation in adaptive information systems. Neural Comput Appl pp. 1–23
Raissi M, Perdikaris P, Karniadakis G (2017) Physics informed deep learning (part ii): Data-driven, discovery of nonlinear partial differential equations,”, arxiv e-prints, p. arXiv preprint arXiv:1711.10566
Raissi M, Perdikaris P, Karniadakis GE (2017) Physics informed deep learning (part i): Data-driven solutions of nonlinear partial differential equations. arXiv preprint arXiv:1711.10561
Raissi M, Perdikaris P, Karniadakis GE (2019) Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. J Comput Phys 378:686–707
Raissi M, Yazdani A, Karniadakis GE (2018) Hidden fluid mechanics: A navier–stokes informed deep learning framework for assimilating flow visualization data. arXiv preprint arXiv:1808.04327
Raissi M, Yazdani A, Karniadakis GE (2020) Hidden fluid mechanics: learning velocity and pressure fields from flow visualizations. Science 367(6481):1026–1030
Ramachandran P, Zoph B, Le QV (2017) Searching for activation functions. arXiv preprint arXiv:1710.05941
Rao C, Sun H, Liu Y (2020) Physics-informed deep learning for incompressible laminar flows. Theor Appl Mech Lett 10(3):207–212
Salimans T, Kingma DP (2016) Weight normalization: a simple reparameterization to accelerate training of deep neural networks. arXiv preprint arXiv:1602.07868
Saurav S, Saini R, Singh S (2021) Emnet: a deep integrated convolutional neural network for facial emotion recognition in the wild. Appl Intell. pp. 1–28
Shams EA, Rizaner A, Ulusoy AH (2021) A novel context-aware feature extraction method for convolutional neural network-based intrusion detection systems. Neural Comput Appl. pp. 1–19
Shanmuganathan V, Yesudhas HR, Khan MS, Khari M, Gandomi AH (2020) R-cnn and wavelet feature extraction for hand gesture recognition with emg signals. Neural Comput Appl 32(21):16723–16736
Shorten C, Khoshgoftaar TM (2019) A survey on image data augmentation for deep learning. J Big Data 6(1):1–48
Subramaniam A, Wong ML, Borker RD, Nimmagadda S, Lele SK (2020) Turbulence enrichment using physics-informed generative adversarial networks. arXiv e-prints pp. arXiv–2003
Wandel N, Weinmann M, Klein R (2020) Fast fluid simulations in 3d with physics-informed deep learning. arXiv preprint arXiv:2012.11893
Xu C, Wang H, Wu S, Lin Z (2021) Treelstm with tag-aware hypernetwork for sentence representation. Neurocomputing 434:11–20
Xu H, Zhang W, Wang Y (2020) Explore missing flow dynamics by physics-informed deep learning: the parameterised governing systems. arXiv preprint arXiv:2008.12266
Yang Z, Wu JL, Xiao H (2019) Enforcing deterministic constraints on generative adversarial networks for emulating physical systems. arXiv preprint arXiv:1911.06671
Yi X, Duan Z, Li R, Zhang J, Li T, Zheng Y (2020) Predicting fine-grained air quality based on deep neural networks. IEEE Transactions on Big Data
Acknowledgements
This work is supported by Natural Science Foundation of Shanghai under Grant 19ZR1417700, Transforming Systems through Partnership, Newton Fund under Grant TSPC1086, High Performance Computing Center of Shanghai University.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Wu, P., Pan, K., Ji, L. et al. Navier–stokes Generative Adversarial Network: a physics-informed deep learning model for fluid flow generation. Neural Comput & Applic 34, 11539–11552 (2022). https://doi.org/10.1007/s00521-022-07042-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-022-07042-6