Title: Graph convolutional networks applied to unstructured flow field data

Journal Article · · Machine Learning: Science and Technology
ORCiD logo; ORCiD logo; ; ORCiD logo

Abstract Many scientific and engineering processes produce spatially unstructured data. However, most data-driven models require a feature matrix that enforces both a set number and order of features for each sample. They thus cannot be easily constructed for an unstructured dataset. Therefore, a graph based data-driven model to perform inference on fields defined on an unstructured mesh, using a graph convolutional neural network (GCNN) is presented. The ability of the method to predict global properties from spatially irregular measurements with high accuracy is demonstrated by predicting the drag force associated with laminar flow around airfoils from scattered velocity measurements. The network can infer from field samples at different resolutions, and is invariant to the order in which the measurements within each sample are presented. The GCNN method, using inductive convolutional layers and adaptive pooling, is able to predict this quantity with a validation R 2 above 0.98, and a Normalized Mean Squared Error below 0.01, without relying on spatial structure.

Sponsoring Organization:
USDOE
OSTI ID:
1835359
Journal Information:
Machine Learning: Science and Technology, Journal Name: Machine Learning: Science and Technology Journal Issue: 4 Vol. 2; ISSN 2632-2153
Publisher:
IOP PublishingCopyright Statement
Country of Publication:
United Kingdom
Language:
English

References (23)

Gmsh: A 3-D finite element mesh generator with built-in pre- and post-processing facilities journal September 2009
A Comparison of Methods for Evaluating Time-Dependent Fluid Dynamic Forces on Bodies, Using only Velocity Fields and Their Derivatives journal July 1999
Dense motion estimation of particle images via a convolutional neural network journal March 2019
Prediction of aerodynamic flow fields using convolutional neural networks journal June 2019
Conservative physics-informed neural networks on discrete domains for conservation laws: Applications to forward and inverse problems journal June 2020
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations journal February 2019
Deep learning observables in computational fluid dynamics journal June 2020
Meshfree methods on manifolds for hydrodynamic flows on curved surfaces: A Generalized Moving Least-Squares (GMLS) approach journal May 2020
nPINNs: Nonlocal physics-informed neural networks for a parametrized nonlocal universal Laplacian operator. Algorithms and applications journal December 2020
Drag coefficient prediction for non-spherical particles in dense gas–solid two-phase flow using artificial neural network journal September 2019
Coherent structure colouring: identification of coherent structures from sparse data using graph theory journal December 2016
An impulse-based approach to estimating forces in unsteady flow journal February 2017
Prediction of turbulent heat transfer using convolutional neural networks journal November 2019
Graph Networks as a Universal Machine Learning Framework for Molecules and Crystals journal April 2019
Solving high-dimensional partial differential equations using deep learning journal August 2018
Performing particle image velocimetry using artificial neural networks: a proof-of-concept journal November 2017
Spectral-clustering approach to Lagrangian vortex detection journal June 2016
Network community-based model reduction for vortical flows journal June 2018
Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties journal April 2018
DOLFIN: Automated finite element computing journal April 2010
Graph Convolutional Encoders for Syntax-aware Neural Machine Translation conference January 2017
GMLS-Nets: A Framework for Learning from Unstructured Data report September 2019
Network-based study of Lagrangian transport and mixing journal January 2017

Similar Records

Related Subjects