Assessment of supervised machine learning methods for fluid flows (2020-01-27T00:00:00.000000Z)

TL;DR

Four machine learning architectures are examined in terms of their characteristics, accuracy, computational cost, and robustness for canonical flow problems, and the use of convolutional neural network is considered in the context of super-resolution analysis of two- dimensional cylinder wake, two-dimensional decaying isotropic turbulence, and three-dimensional turbulent channel flow.

Abstract

We apply supervised machine learning techniques to a number of regression problems in fluid dynamics. Four machine learning architectures are examined in terms of their characteristics, accuracy, computational cost, and robustness for canonical flow problems. We consider the estimation of force coefficients and wakes from a limited number of sensors on the surface for flows over a cylinder and NACA0012 airfoil with a Gurney flap. The influence of the temporal density of the training data is also examined. Furthermore, we consider the use of convolutional neural network in the context of super-resolution analysis of two-dimensional cylinder wake, two-dimensional decaying isotropic turbulence, and three-dimensional turbulent channel flow. In the concluding remarks, we summarize on findings from a range of regression-type problems considered herein.

Authors

Kai Fukami

1 papers

K. Fukagata

1 papers

K. Taira

1 papers

TL;DR

Abstract

Authors

References80 items

Perspective on machine learning for advancing fluid mechanics

Time-resolved turbulent velocity field reconstruction using a long short-term memory (LSTM)-based artificial intelligence framework

Time-series learning of latent-space dynamics for reduced-order model closure

Nonlinear mode decomposition with convolutional neural networks for fluid dynamics

Machine Learning for Fluid Mechanics

Predictions of turbulent shear flows using deep neural networks

Data-Driven Science and Engineering

Modal Analysis of Fluid Flows: Applications and Outlook

Construction of reduced-order models for fluid flows using deep feedforward neural networks

Revealing essential dynamics from high-dimensional fluid flow data and operators

Shallow Learning for Fluid Flow Reconstruction with Limited Sensors and Limited Data

Flowfield Reconstruction Method Using Artificial Neural Network

Mixing and reaction in turbulent plumes: the limits of slow and instantaneous chemical kinetics

Recovering missing CFD data for high-order discretizations using deep neural networks and dynamics learning

Super-resolution reconstruction of turbulent flows with machine learning

Deep learning of mixing by two ‘atoms’ of stratified turbulence

Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control

Subgrid modelling for two-dimensional turbulence using neural networks

Synthetic turbulent inflow generator using machine learning

A Novel Deep Learning Method for the Predictions of Current Forces on Bluff Bodies

Network community-based model reduction for vortical flows.

Inferring flow parameters and turbulent configuration with physics-informed data assimilation and spectral nudging

Unraveling turbulence via physics-informed data-assimilation and spectral nudging

Turbulence Modeling in the Age of Data

Extreme learning machine for reduced order modeling of turbulent geophysical flows.

Single Image Super-Resolution Based on Multi-Scale Competitive Convolutional Neural Network

Application of Convolutional Neural Network to Predict Airfoil Lift Coefficient

An Implantable Wireless Neural Interface System for Simultaneous Recording and Stimulation of Peripheral Nerve with a Single Cuff Electrode

Airfoil wake modification with Gurney flap at Low-Reynolds number

A neural network approach for the blind deconvolution of turbulent flows

Modal Analysis of Fluid Flows: An Overview

Deep learning in fluid dynamics

Machine Learning Control – Taming Nonlinear Dynamics and Turbulence

Reynolds averaged turbulence modelling using deep neural networks with embedded invariance

Example-Based Super Resolution

Searching for turbulence models by artificial neural network

Network structure of two-dimensional decaying isotropic turbulence

Linear stability and energetics of rotating radial horizontal convection

Evaluation of machine learning algorithms for prediction of regions of high Reynolds averaged Navier Stokes uncertainty

Adam: A Method for Stochastic Optimization

Making a Science of Model Search: Hyperparameter Optimization in Hundreds of Dimensions for Vision Architectures

A few useful things to know about machine learning

Regression and the Moore-Penrose Pseudoinverse

A Tutorial on Bayesian Optimization of Expensive Cost Functions, with Application to Active User Modeling and Hierarchical Reinforcement Learning

Rectified Linear Units Improve Restricted Boltzmann Machines

Iterative-Interpolation Super-Resolution Image Reconstruction - A Computationally Efficient Technique

A fast immersed boundary method using a nullspace approach and multi-domain far-field boundary conditions

Decision Tree Instability and Active Learning

The immersed boundary method: A projection approach

A theoretical prediction of friction drag reduction in turbulent flow by superhydrophobic surfaces

A tutorial on support vector regression

Extreme learning machine: a new learning scheme of feedforward neural networks

Matrices: Theory and Applications

Popular Ensemble Methods: An Empirical Study

Neural Networks: A Comprehensive Foundation

Effect of data standardization on neural network training

Arbitrary nonlinearity is sufficient to represent all functions by neural networks: A theorem

Approximation capabilities of multilayer feedforward networks

Approximation by superpositions of a sigmoidal function

Learning representations by back-propagating errors

A new type of host compound consisting of α-zirconium phosphate and an animated cyclodextrin

Machine learning, dynamical systems, and control, Cambridge Univ

Why Deep Neural Networks: A Possible Theoretical Explanation

Pattern Recognition And Machine Learning

Dropout: a simple way to prevent neural networks from overfitting

SpringerVerlag

Induction of decision trees

Random Forests

Haykin, Simon. Neural networks: A comprehensive foundation, Prentice Hall, Inc. Segunda Edición, 1999

An empirical study, J

Gradient-based learning applied to document recognition

Learning representations of back-propagation errors

Some methods for classification and analysis of multivariate observations

Some methods for classi cation and analysis of multivariate observations

Pattern recognition using generalized portrait method

Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations