Physics-Driven Learning of the Steady Navier-Stokes Equations using Deep Convolutional Neural Networks
From MaRDI portal
Publication:5042008
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.
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
Cites work
- A paradigm for data-driven predictive modeling using field inversion and machine learning
- Deep learning
- Gmsh: a 3-D finite element mesh generator with built-in pre- and post-processing facilities
- Hidden physics models: machine learning of nonlinear partial differential equations
- scientific article; zbMATH DE number 1015078 (Why is no real title available?)
- Model reduction for compressible flows using POD and Galerkin projection
- Non-intrusive reduced order modeling of nonlinear problems using neural networks
- Non-intrusive reduced order modeling of unsteady flows using artificial neural networks with application to a combustion problem
- Numerical simulation of laminar flow past a circular cylinder
- PDE-Net 2.0: learning PDEs from data with a numeric-symbolic hybrid deep network
- 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
- Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
- Turbulence modeling in the age of data
Cited in
(33)- Discretizationnet: a machine-learning based solver for Navier-Stokes equations using finite volume discretization
- Deep learning of the spanwise-averaged Navier-Stokes equations
- 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
- DeepM\&Mnet: inferring the electroconvection multiphysics fields based on operator approximation by neural networks
- Mesh-Conv: convolution operator with mesh resolution independence for flow field modeling
- Error estimates for deep learning methods in fluid dynamics
- Physics-informed PointNet: a deep learning solver for steady-state incompressible flows and thermal fields on multiple sets of irregular geometries
- Data-driven modelling of nonlinear spatio-temporal fluid flows using a deep convolutional generative adversarial network
- Physics informed by deep learning: numerical solutions of modified Korteweg-de Vries equation
- ReF-nets: physics-informed neural network for Reynolds equation of gas bearing
- Structure preservation for the deep neural network multigrid solver
- Surrogate convolutional neural network models for steady computational fluid dynamics simulations
- 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
- Flow over an espresso cup: inferring 3-D velocity and pressure fields from tomographic background oriented Schlieren via physics-informed neural networks
- Deep learning of vortex-induced vibrations
- A tailored convolutional neural network for nonlinear manifold learning of computational physics data using unstructured spatial discretizations
- Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations
- Learned turbulence modelling with differentiable fluid solvers: physics-based loss functions and optimisation horizons
- Stationary flow predictions using convolutional neural networks
- Learning to Discretize: Solving 1D Scalar Conservation Laws via Deep Reinforcement Learning
- Deep learning in turbulent convection networks
- Learning Invariant Representation of Multiscale Hyperelastic Constitutive Law from Sparse Experimental Data
- Physics-informed neural networks for the Reynolds-averaged Navier-Stokes modeling of Rayleigh-Taylor turbulent mixing
- Linearized Learning with Multiscale Deep Neural Networks for Stationary Navier-Stokes Equations with Oscillatory Solutions
- Neural vortex method: from finite Lagrangian particles to infinite dimensional Eulerian dynamics
- Less Emphasis on Hard Regions: Curriculum Learning of PINNs for Singularly Perturbed Convection-Diffusion-Reaction Problems
- Splitting physics-informed neural networks for inferring the dynamics of integer- and fractional-order neuron models
- Machine learning algorithm for the Monge-Ampère equation with transport boundary conditions
- Construction of an artificial neural network for solving the incompressible Navier-Stokes equations
- MHDnet: physics-preserving learning for solving magnetohydrodynamics problems
- Multi-artificial neural network for inverse eigenvalue problem with the weighted Helmholtz equation
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)
