A novel spatial-temporal prediction method for unsteady wake flows based on hybrid deep neural network
Year: 2019
Type: article
Abstract: A fast and accurate prediction method of unsteady flow is a challenge in fluid dynamics due to the high-dimensional and nonlinear dynamic behavior. A novel hybrid deep neural network (DNN) architecture was designed to capture the spatial-temporal features of unsteady flows directly from high-dimensional numerical unsteady flow field data. The hybrid DNN is constituted by the convolutional neural network, convolutional long short term memory neural network, and deconvolutional neural network. The unsteady wake flow around a cylinder at various Reynolds numbers and an airfoil at a higher Reynolds number are calculated to establish the datasets as training samples of the hybrid DNN. The trained hybrid DNNs were then tested by predicting the unsteady flow fields in future time steps. The predicted flow fields using the trained hybrid DNN are in good agreement with those calculated directly by a computational fluid dynamic solver.
Source: Physics of Fluids
DOI: 10.1063/1.5127247
Authors: Renkun Han, Yixing Wang, Yang Zhang, Gang Chen
Language: English
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2965300934