Deep learning discovery of macroscopic governing equations for viscous gravity currents from microscopic simulation data
From MaRDI portal
Publication:6181709
Abstract: Although deep-learning has been successfully applied in a variety of science and engineering problems owing to its strong high-dimensional nonlinear mapping capability, it is of limited use in scientific knowledge discovery. In this work, we propose a deep-learning based framework to discover the macroscopic governing equation of viscous gravity current based on high-resolution microscopic simulation data without the need for prior knowledge of underlying terms. For two typical scenarios with different viscosity ratios, the deep-learning based equations exactly capture the same dominated terms as the theoretically derived equations for describing long-term asymptotic behaviors, which validates the proposed framework. Unknown macroscopic equations are then obtained for describing short-term behaviors, and additional deep-learned compensation terms are eventually discovered. Comparison of posterior tests shows that the deep-learning based PDEs actually perform better than the theoretically derived PDEs in predicting evolving viscous gravity currents for both long-term and short-term regimes. Moreover, the proposed framework is proven to be very robust against non-biased data noise for training, which is up to 20%. Consequently, the presented deep-learning framework shows considerable potential for discovering unrevealed intrinsic laws in scientific semantic space from raw experimental or simulation results in data space.
Recommendations
- Deep hidden physics models: deep learning of nonlinear partial differential equations
- DeepMoD: deep learning for model discovery in noisy data
- DL-PDE: Deep-Learning Based Data-Driven Discovery of Partial Differential Equations from Discrete and Noisy Data
- A physics-informed operator regression framework for extracting data-driven continuum models
- Data-driven discovery of PDEs in complex datasets
Cites work
- A conservative level set method for two phase flow
- A dipole solution for power-law gravity currents in porous formations
- An aggregation-based algebraic multigrid method
- DL-PDE: Deep-Learning Based Data-Driven Discovery of Partial Differential Equations from Discrete and Noisy Data
- DLGA-PDE: discovery of PDEs with incomplete candidate library via combination of deep learning and genetic algorithm
- Data-driven discovery of PDEs in complex datasets
- Data-driven identification of parametric partial differential equations
- Dispersion induced by non-Newtonian gravity flow in a layered fracture or formation
- Draining viscous gravity currents in a vertical fracture
- Dynamics of fluids in porous media.
- Efficient implementation of weighted ENO schemes
- Gravity-driven flow of Herschel-Bulkley fluid in a fracture and in a 2D porous medium
- Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations
- Influence of heterogeneity on second-kind self-similar solutions for viscous gravity currents
- Machine learning of linear differential equations using Gaussian processes
- Machine learning subsurface flow equations from data
- On the slow draining of a gravity current moving through a layered permeable medium
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- Robust data-driven discovery of governing physical laws with error bars
- Sparse identification of nonlinear dynamics for model predictive control in the low-data limit
- Spreading and imbibition of viscous liquid on a porous base. II
- The dynamics of two-layer gravity-driven flows in permeable rock
- Total variation diminishing Runge-Kutta schemes
This page was built for publication: Deep learning discovery of macroscopic governing equations for viscous gravity currents from microscopic simulation data
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6181709)
