Shallow neural networks for fluid flow reconstruction with limited sensors

DOI10.1098/RSPA.2020.0097zbMATH Open1472.68172arXiv1902.07358OpenAlexW3035295279WikidataQ98649521 ScholiaQ98649521MaRDI QIDQ5160984FDOQ5160984

Authors: N. Benjamin Erichson, L. Mathelin, Zhewei Yao, S. L. Brunton, Michael W. Mahoney, J. N. Kutz 

Publication date: 29 October 2021

Published in: Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences (Search for Journal in Brave)

Abstract: In many applications, it is important to reconstruct a fluid flow field, or some other high-dimensional state, from limited measurements and limited data. In this work, we propose a shallow neural network-based learning methodology for such fluid flow reconstruction. Our approach learns an end-to-end mapping between the sensor measurements and the high-dimensional fluid flow field, without any heavy preprocessing on the raw data. No prior knowledge is assumed to be available, and the estimation method is purely data-driven. We demonstrate the performance on three examples in fluid mechanics and oceanography, showing that this modern data-driven approach outperforms traditional modal approximation techniques which are commonly used for flow reconstruction. Not only does the proposed method show superior performance characteristics, it can also produce a comparable level of performance with traditional methods in the area, using significantly fewer sensors. Thus, the mathematical architecture is ideal for emerging global monitoring technologies where measurement data are often limited.

Full work available at URL: https://arxiv.org/abs/1902.07358

Recommendations

• State estimation with limited sensors -- a deep learning based approach
• Leveraging reduced-order models for state estimation using deep learning
• Unsteady flow prediction from sparse measurements by compressed sensing reduced order modeling
• Linearly recurrent autoencoder networks for learning dynamics
• Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations

zbMATH Keywords

machine learningartificial intelligenceneural networksfluid dynamicssensorscomputational physicsmechanical engineeringflow field estimation

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Basic methods in fluid mechanics (76M99)

Cites Work

• Sparse principal component analysis via variable projection
• Title not available (Why is that?)
• Title not available (Why is that?)
• Regularization and Variable Selection Via the Elastic Net
• Acceleration of Stochastic Approximation by Averaging
• Finding structure with randomness: probabilistic algorithms for constructing approximate matrix decompositions
• A Statistical View of Some Chemometrics Regression Tools
• A fast immersed boundary method using a nullspace approach and multi-domain far-field boundary conditions
• The immersed boundary method: a projection approach
• Robust uncertainty principles: exact signal reconstruction from highly incomplete frequency information
• A hierarchy of low-dimensional models for the transient and post-transient cylinder wake
• Compressed sensing
• Randomized Algorithms for Matrices and Data
• Data-Driven Sparse Sensor Placement for Reconstruction: Demonstrating the Benefits of Exploiting Known Patterns
• An `empirical interpolation' method: Application to efficient reduced-basis discretization of partial differential equations
• Nonlinear model reduction via discrete empirical interpolation
• Unsteady flow sensing and estimation via the gappy proper orthogonal decomposition
• A survey of projection-based model reduction methods for parametric dynamical systems
• Image Super-Resolution Via Sparse Representation
• Model Reduction for Flow Analysis and Control
• A public turbulence database cluster and applications to study Lagrangian evolution of velocity increments in turbulence
• Hidden physics models: machine learning of nonlinear partial differential equations
• Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks
• Randomized Dynamic Mode Decomposition
• Reynolds averaged turbulence modelling using deep neural networks with embedded invariance
• Solving ill-posed inverse problems using iterative deep neural networks
• Deep Convolutional Neural Network for Inverse Problems in Imaging
• Deep convolutional framelets: a general deep learning framework for inverse problems
• Reconciling modern machine-learning practice and the classical bias-variance trade-off
• Image restoration using a neural network
• Observable dictionary learning for high-dimensional statistical inference
• A new selection operator for the discrete empirical interpolation method -- improved a priori error bound and extensions
• Recovering missing CFD data for high-order discretizations using deep neural networks and dynamics learning
• Machine-learning-based spatio-temporal super resolution reconstruction of turbulent flows

Cited In (16)

• A causation-based computationally efficient strategy for deploying Lagrangian drifters to improve real-time state estimation
• Optimal sensor placement for ensemble-based data assimilation using gradient-weighted class activation mapping
• State estimation with limited sensors -- a deep learning based approach
• Reconstruction of three-dimensional turbulent flow structures using surface measurements for free-surface flows based on a convolutional neural network
• Data-driven transient lift attenuation for extreme vortex gust-airfoil interactions
• Complete flow characterization from snapshot PIV, fast probes and physics-informed neural networks
• Discrepancy Modeling Framework: Learning Missing Physics, Modeling Systematic Residuals, and Disambiguating between Deterministic and Random Effects
• Launching Drifter Observations in the Presence of Uncertainty
• Physics-informed ConvNet: learning physical field from a shallow neural network
• Data-driven stochastic particle scheme for collisional plasma simulations
• Informative and non-informative decomposition of turbulent flow fields
• Nonlinear mode decomposition with convolutional neural networks for fluid dynamics
• Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression
• Real-time estimation and prediction of unsteady flows using reduced-order models coupled with few measurements
• Gappy AE: a nonlinear approach for gappy data reconstruction using auto-encoder
• Leveraging reduced-order models for state estimation using deep learning

This page was built for publication: Shallow neural networks for fluid flow reconstruction with limited sensors

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5160984)