Deep Learning-Based Prediction of Aerodynamic Performance for Airfoils in Transonic Regime

This paper presents an approach to estimate the aerodynamic coefficients of airfoils in the transonic regime using Artificial Neural Networks. The transonic regime is a critical and challenging aerodynamic domain, and our approach utilizes data generated by the OpenFOAM
^®
to train our model. Our dataset encompasses a wide range of transonic flow conditions and different airfoil shapes, enabling our Artificial Neural Networks to capture the complex behavior of aerodynamic phenomena in this regime. Our proposed framework achieves high accuracy, with the lift and moment coefficient predictions demonstrating an unprecedented accuracy level of 99.7% with respect to the test dataset obtained by OpenFOAM
^®
. Our results demonstrate the potential of Artificial Neural Networks to accurately predict aerodynamic coefficients in the transonic regime, which could have significant implications for the design and optimization of high-performance aircraft.