Abstract:During aerodynamic design, Computational Fluid Dynamics (CFD) can perform high-precision calculations regarding different flow conditions, but it takes too long to solve the large-scale and complex partial differential equations which makes it difficult for designers to fully explore the design scope. With the rapid development of deep learning methods specifically applied to the Convolutional Neural Networks (CNN), researchers have designed some image-based CNN flow field prediction models. However, the traditional CNN cannot process the non-Euclidean spatial flow field problems, and the end-to-end methods often cannot guarantee the validity of the data. In this paper, we propose a Graph Convolutional Network (GCN) flow field prediction network model coupled with a flow field solver FlowGCN, in which the GCN can be effectively integrated with the computational grids and flow conditions, and combined with the traditional numerical solver to ensure the validity of the simulation results. The core of FlowGCN is, by predicting the physical quantities of the flow field through the GCN, such as the velocity field, pressure field and eddy viscosity, to provide the solver with an initial flow field closer to the convergence state, thereby reducing the number of iterations of the numerical solution. In the steady flow model, the typical two-dimensional airfoil calculation example shows that FlowGCN can predict various physical field quantities well under different flow conditions. In addition, because the graph network directly corresponds to the calculation grids, the lossless predicted flow field can be used as the initial field of the solver, and the more reliable simulation results can be calculated by the solver. Compared with the direct calculation by the solver, it can get 1.9 times faster.

Swin-FlowNet: Flow field oriented optimization aided by a CNN and Swin-Transformer based model

Fast Aerodynamic Prediction of Airfoil with Trailing Edge Flap Based on Multi-Task Deep Learning

An Accuracy-Enhanced Transonic Flow Prediction Method Fusing Deep Learning and a Reduced-Order Model

A Swin-Transformer-based Model for Efficient Compression of Turbulent Flow Data

Sag-flownet: self-attention generative network for airfoil flow field prediction

Flow Field Modeling of Airfoil Based on Convolutional Neural Networks from Transform Domain Perspective

Fast Transonic Flow Prediction Enables Efficient Aerodynamic Design

Fast simulation of airfoil flow field via deep neural network

A Novel Deep Learning Model for the Flow Field Reconstruction of An Oscillating Airfoil

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

Developing an advanced neural network and physics solver coupled framework for accelerating flow field simulations

A deep learning approach for hydrofoil optimization of tidal turbines

A Physics-Constrained and Data-Driven Method for Modeling Supersonic Flow

Fast Flow Field Prediction of Hydrofoils Based on Deep Learning

A solver for subsonic flow around airfoils based on physics-informed neural networks and mesh transformation

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

Research on the Role of Hybrid Mesh Warm-up in Flow Prediction Based on Deep Learning

Prediction of Aerodynamic Flow Fields Using Convolutional Neural Networks