Flowfield Prediction of Airfoil Off-Design Conditions Based on a Modified Variational Autoencoder

Airfoil aerodynamic optimization based on single-point design may lead to poor off-design behaviors. Multipoint that considers the flow conditions is usually applied improve robustness and expand flight envelope. Many deep learning models have been utilized for rapid prediction or reconstruction of flowfields. However, flowfield accuracy be insufficient cruise efficiency optimization, model generalization ability also questionable when facing airfoils different from with which has trained. Because a computational fluid dynamic evaluation condition necessary affordable in industrial design, novel framework proposed utilize as prior reference prediction. A variational autoencoder developed extract features generate new flowfields under other free stream conditions. Physical-based loss functions force conservation mass are derived minimize error reconstruction. The results demonstrate can reduce test by 30% compared traditional models. physical-based function further 4%. illustrates better balance time cost fidelity requirements conditions, makes more feasible applications.