Model order reduction with neural networks: Application to laminar and turbulent flows

TL;DR

This paper explores neural network-based model order reduction using autoencoders for various fluid flows, assessing their effectiveness and parameter sensitivity across laminar and turbulent regimes.

Contribution

It demonstrates the application of convolutional autoencoders to fluid flow data and analyzes their sensitivity to model parameters for different flow types.

Findings

Autoencoders can effectively reduce fluid flow models.

Model performance depends on parameter choices like latent modes and activation functions.

Neural network-based reduction shows promise for fluid dynamics applications.

Abstract

We investigate the capability of neural network-based model order reduction, i.e., autoencoder (AE), for fluid flows. As an example model, an AE which comprises of a convolutional neural network and multi-layer perceptrons is considered in this study. The AE model is assessed with four canonical fluid flows, namely: (1) two-dimensional cylinder wake, (2) its transient process, (3) NOAA sea surface temperature, and (4) $y-z$ sectional field of turbulent channel flow, in terms of a number of latent modes, a choice of nonlinear activation functions, and a number of weights contained in the AE model. We find that the AE models are sensitive against the choice of the aforementioned parameters depending on the target flows. Finally, we foresee the extensional applications and perspectives of machine learning based order reduction for numerical and experimental studies in fluid dynamics community.