Deep neural networks for fast aerodynamic predictions

This work proposes a novel multi-output neural network for the prediction of the aerodynamic coefficients of wings in three dimensions using inviscid compressible flow data. Contrary to existing neural networks that are designed to predict the aerodynamic coefficients directly, the proposed network considers as output the pressure at a number of selected points on the aerodynamic shape. The performance of the proposed neural network is compared against the existing neural networks. The numerical results, involving high dimensional problems with flow and geometric parameters, show the benefits of the proposed approach.