Abstract:In this article, we propose an unsteady data-driven reduced order model (ROM) (surrogate model) for predicting the velocity field around an airfoil. The network model applies a convolutional neural network (CNN) as the encoder and a deconvolutional neural network (DCNN) as the decoder. The model constructs a mapping function between temporal evolution of the pressure signal on the airfoil surface and the surrounding velocity field. For improving the model performance, the input matrix is designed to further incorporate the information of the Reynolds number, the geometry of the airfoil, and the angle of attack. The DCNN works as the decoder for better reconstructing the spatial and temporal information of the features extracted by the CNN encoder. The training and testing datasets of flow fields under different conditions are obtained by solving the Navier–Stokes equations using the computational fluid dynamics method. After model training, the neural network based ROM shows accurate and dramatically fast predictions on the flow field of the testing dataset with extended angles of attack and Reynolds numbers. According to the current study, the neural network-based ROM has exhibited attractive potentials on ROM of the unsteady fluid dynamic problem, and the model can potentially serve on investigating flow control or optimization problems in the future.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to use deep neural networks to predict the time - varying flow field around the wing under different Reynolds numbers and angles of attack. Specifically, the authors propose a data - driven reduced - order model (ROM) based on deep learning. This model uses a convolutional neural network (CNN) as an encoder and a deconvolutional neural network (DCNN) as a decoder to construct the mapping relationship between the time evolution of the pressure signal on the wing surface and the surrounding flow velocity field. Through this method, the model can quickly and accurately predict the flow field under different flow conditions, which requires a large amount of time and computational resources in the traditional computational fluid dynamics (CFD) method. The main contributions of the paper are as follows: 1. **Improve prediction speed**: Compared with the traditional CFD method, the proposed neural network model can significantly accelerate the prediction speed of the flow field, which is especially important for optimization and flow control problems that require a large number of simulations. 2. **Expand the application range**: The model not only performs well within the training data range, but also can effectively predict the flow field under Reynolds numbers and angles of attack conditions outside the training data range, demonstrating the generalization ability of the model. 3. **Retain spatial information**: By using DCNN as a decoder, the model can better reconstruct the spatial and temporal information of the flow field and avoid the possible loss of spatial information caused by the traditional fully - connected layer. These features make the model have great potential in future fluid dynamics research and practical applications, especially in fields such as flight control and flow field optimization.

Time-variant prediction of flow over an airfoil using deep neural network

Reduced-order modeling of unsteady fluid flow using neural network ensembles

Geometry and boundary condition adaptive data-driven model of fluid flow based on deep convolutional neural networks

Fast simulation of airfoil flow field via deep neural network

Data-driven reduced-order modeling for nonlinear aerodynamics using an autoencoder neural network

Hybrid deep neural network based prediction method for unsteady flows with moving boundary

Prediction of Aerodynamic Flow Fields Using Convolutional Neural Networks

Deep neural networks for nonlinear model order reduction of unsteady flows

An accuracy-enhanced transonic flow prediction method fusing deep learning and reduced-order model

Fast Transonic Flow Prediction Enables Efficient Aerodynamic Design

Fast fluid–structure interaction simulation method based on deep learning flow field modeling

Deep learning-based predictive modelling of transonic flow over an aerofoil

Reduced-Order Modeling of Steady and Unsteady Flows with Deep Neural Networks

Unsteady reduced-order model of flow over cylinders based on convolutional and deconvolutional neural network structure

Prediction of transonic flow over supercritical airfoils using geometric-encoding and deep-learning strategies

A Non-Intrusive Reduced Order Model with Transformer Neural Network and Its Application

Application of Deep Learning Method to Reynolds Stress Models of Channel Flow Based on Reduced-Order Modeling of DNS Data

Reduced Order Data-Driven Analysis of Cavitating Flow over Hydrofoil with Machine Learning

An improved neural network for modeling airfoil's unsteady aerodynamics in transonic flow

An enhanced hybrid deep neural network reduced-order model for transonic buffet flow prediction