Physics-Driven Learning of the Steady Navier-Stokes Equations using Deep Convolutional Neural Networks
From MaRDI portal
Publication:5042008
DOI10.4208/CICP.OA-2021-0146zbMATH Open1500.76071arXiv2106.09301OpenAlexW3167808783WikidataQ114911802 ScholiaQ114911802MaRDI QIDQ5042008FDOQ5042008
Authors: Hao Ma, Yuxuan Zhang, Nils Thuerey, Xiangyu Hu, Oskar J. Haidn 
Publication date: 18 October 2022
Published in: Communications in Computational Physics (Search for Journal in Brave)
Abstract: Recently, physics-driven deep learning methods have shown particular promise for the prediction of physical fields, especially to reduce the dependency on large amounts of pre-computed training data. In this work, we target the physics-driven learning of complex flow fields with high resolutions. We propose the use of emph{Convolutional neural networks} (CNN) based U-net architectures to efficiently represent and reconstruct the input and output fields, respectively. By introducing Navier-Stokes equations and boundary conditions into loss functions, the physics-driven CNN is designed to predict corresponding steady flow fields directly. In particular, this prevents many of the difficulties associated with approaches employing fully connected neural networks. Several numerical experiments are conducted to investigate the behavior of the CNN approach, and the results indicate that a first-order accuracy has been achieved. Specifically for the case of a flow around a cylinder, different flow regimes can be learned and the adhered "twin-vortices" are predicted correctly. The numerical results also show that the training for multiple cases is accelerated significantly, especially for the difficult cases at low Reynolds numbers, and when limited reference solutions are used as supplementary learning targets.
Full work available at URL: https://arxiv.org/abs/2106.09301
Recommendations
- Deep learning of the spanwise-averaged Navier-Stokes equations
- Deep learning observables in computational fluid dynamics
- Learning by neural networks under physical constraints for simulation in fluid mechanics
- The deep learning Galerkin method for the general Stokes equations
- Linearized Learning with Multiscale Deep Neural Networks for Stationary Navier-Stokes Equations with Oscillatory Solutions
- Active training of physics-informed neural networks to aggregate and interpolate parametric solutions to the Navier-Stokes equations
- NSFnets (Navier-Stokes flow nets): physics-informed neural networks for the incompressible Navier-Stokes equations
- Physics-informed neural networks with parameter asymptotic strategy for learning singularly perturbed convection-dominated problem
- Deep learning in fluid dynamics
- Surrogate convolutional neural network models for steady computational fluid dynamics simulations
incompressible Navier-Stokes equationslow Reynolds number flowfirst-order accuracyiterative stochastic gradient descent methodU-net architecture
Learning and adaptive systems in artificial intelligence (68T05) Navier-Stokes equations for incompressible viscous fluids (76D05) Basic methods in fluid mechanics (76M99)
Cites Work
- Gmsh: a 3-D finite element mesh generator with built-in pre- and post-processing facilities
- PDE-Net 2.0: learning PDEs from data with a numeric-symbolic hybrid deep network
- Title not available (Why is that?)
- Title not available (Why is that?)
- Model reduction for compressible flows using POD and Galerkin projection
- Numerical simulation of laminar flow past a circular cylinder
- A paradigm for data-driven predictive modeling using field inversion and machine learning
- Turbulence Modeling in the Age of Data
- Non-intrusive reduced order modeling of nonlinear problems using neural networks
- Hidden physics models: machine learning of nonlinear partial differential equations
- Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- Reynolds averaged turbulence modelling using deep neural networks with embedded invariance
- Non-intrusive reduced order modeling of unsteady flows using artificial neural networks with application to a combustion problem
- Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
Cited In (14)
- Physics-informed neural networks for the Reynolds-averaged Navier-Stokes modeling of Rayleigh-Taylor turbulent mixing
- Physics informed by deep learning: numerical solutions of modified Korteweg-de Vries equation
- Learning to Discretize: Solving 1D Scalar Conservation Laws via Deep Reinforcement Learning
- Discretizationnet: a machine-learning based solver for Navier-Stokes equations using finite volume discretization
- NSFnets (Navier-Stokes flow nets): physics-informed neural networks for the incompressible Navier-Stokes equations
- Less Emphasis on Hard Regions: Curriculum Learning of PINNs for Singularly Perturbed Convection-Diffusion-Reaction Problems
- Machine learning algorithm for the Monge-Ampère equation with transport boundary conditions
- Learned turbulence modelling with differentiable fluid solvers: physics-based loss functions and optimisation horizons
- Deep learning in turbulent convection networks
- Learning Invariant Representation of Multiscale Hyperelastic Constitutive Law from Sparse Experimental Data
- Neural vortex method: from finite Lagrangian particles to infinite dimensional Eulerian dynamics
- Bounded nonlinear forecasts of partially observed geophysical systems with physics-constrained deep learning
- Theoretical prerequisites for physically justified machine learning and its applications to fluid dynamics
- Splitting physics-informed neural networks for inferring the dynamics of integer- and fractional-order neuron models
Uses Software
This page was built for publication: Physics-Driven Learning of the Steady Navier-Stokes Equations using Deep Convolutional Neural Networks
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5042008)
