An automatic shape-aware method for predicting heat flux of supersonic aircraft based on a deep learning approach

Sustained hypersonic flight within the atmosphere causes aerodynamic heating, which poses huge challenges for the thermal protection systems of hypersonic aircraft. Therefore, the heat flux on the aircraft surface needs to be computed accurately during the aircraft design stage. Previous approaches have not been able to achieve simultaneous accuracy and efficiency when computing the heat flux. To deal with this problem, an efficient heat flux prediction method based on deep learning techniques, called SA-HFNet, is proposed in this paper. SA-HFNet tries to learn the relationship between the heat flux and the aircraft shape and flight conditions using deep neural networks without solving the Navier–Stokes equations. Unlike other intelligent methods, SA-HFNet can automatically become aware of changes in aircraft shape. As far as we know, it is the first intelligent method that is able to obtain the heat flux quickly and adapt to changes both in the global aircraft shape and in local shape deformation. Extensive experimental results show that SA-HFNet achieves promising prediction accuracy in less time compared with computational fluid dynamics methods. Furthermore, SA-HFNet has good generalization capability because it has the potential to predict the heat flux for previously unseen aircraft shapes.