Interpreting neural network models of residual scalar flux
From MaRDI portal
(Redirected from Publication:5144568)
Abstract: We show that in addition to providing effective and competitive closures, when analysed in terms of dynamics and physically-relevant diagnostics, artificial neural networks (ANNs) can be both interpretable and provide useful insights in the on-going task of developing and improving turbulence closures. In the context of large-eddy simulations (LES) of a passive scalar in homogeneous isotropic turbulence, exact subfilter fluxes obtained by filtering direct numerical simulations (DNS) are used both to train deep ANN models as a function of filtered variables, and to optimise the coefficients of a turbulent Prandtl number LES closure. extit{A-priori} analysis of the subfilter scalar variance transfer rate demonstrates that learnt ANN models out-perform optimised turbulent Prandtl number closures and Clark-type gradient models. Next, extit{a-posteriori} solutions are obtained with each model over several integral timescales. These experiments reveal, with single- and multi-point diagnostics, that ANN models temporally track exact resolved scalar variance with greater accuracy compared to other subfilter flux models for a given filter length scale. Finally, we interpret the artificial neural networks statistically with differential sensitivity analysis to show that the ANN models feature dynamics reminiscent of so-called "mixed models", where mixed models are understood as comprising both a structural and functional component. Besides enabling enhanced-accuracy LES of passive scalars henceforth, we anticipate this work to contribute to utilising neural network models as a tool in interpretability, robustness and model discovery.
Recommendations
- Subgrid-scale model for large-eddy simulation of isotropic turbulent flows using an artificial neural network
- A machine learning framework for LES closure terms
- A priori sub-grid modelling using artificial neural networks
- Artificial-neural-network-based nonlinear algebraic models for large-eddy simulation of compressible wall-bounded turbulence
- Neural networks based subgrid scale modeling in large eddy simulations
Cites work
- A deterministic forcing scheme for direct numerical simulations of turbulence
- A modulated gradient model for large-eddy simulation: Application to a neutral atmospheric boundary layer
- A neural network approach for the blind deconvolution of turbulent flows
- An approximate deconvolution procedure for large-eddy simulation
- Approximation by superpositions of a sigmoidal function
- Area of scalar isosurfaces in homogeneous isotropic turbulence as a function of Reynolds and Schmidt numbers
- Deep learning in fluid dynamics
- Deep learning of mixing by two `atoms' of stratified turbulence
- Direct numerical simulation of a passive scalar with imposed mean gradient in isotropic turbulence
- Discovering governing equations from data by sparse identification of nonlinear dynamical systems
- Evaluation of subgrid-scale models using an accurately simulated turbulent flow
- Galilean invariance of subgrid-scale stress models in the large-eddy simulation of turbulence
- Inferring solutions of differential equations using noisy multi-fidelity data
- Large Eddy Simulation for Incompressible Flows
- Machine learning for fluid mechanics
- Neural networks based subgrid scale modeling in large eddy simulations
- Optimal LES formulations for isotropic turbulence
- Passive Scalars in Turbulent Flows
- Reynolds averaged turbulence modelling using deep neural networks with embedded invariance
- Statistics of turbulence subgrid-scale stresses: Necessary conditions and experimental tests
- Sub-grid scale model classification and blending through deep learning
- Subgrid modelling for two-dimensional turbulence using neural networks
Cited in
(8)- Data-driven fractional subgrid-scale modeling for scalar turbulence: a nonlocal LES approach
- Stable \textit{a posteriori} LES of 2D turbulence using convolutional neural networks: backscattering analysis and generalization to higher \(Re\) via transfer learning
- A machine learning framework for LES closure terms
- On the role of vorticity stretching and strain self-amplification in the turbulence energy cascade
- Tempered fractional LES modeling
- A physics-inspired alternative to spatial filtering for large-eddy simulations of turbulent flows
- Feature importance in neural networks as a means of interpretation for data-driven turbulence models
- A data-driven method for modelling dissipation rates in stratified turbulence
This page was built for publication: Interpreting neural network models of residual scalar flux
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5144568)
