Physics of Fluids, Год журнала: 2024, Номер 36(11)
Опубликована: Ноя. 1, 2024
This paper proposes a super-resolution (SR) reconstruction method based on deep learning, which efficiently reconstructs the global high-resolution wake flow field from low-resolution (LR) data of propeller. The extensive for propeller under various operating conditions are generated using numerical simulations delayed detached eddy simulation model. proposed approach, convolutional neural networks (PSCNN), uses dilated module to capture multi-scale spatial characteristics fields. performance SR is evaluated by improving resolution different scaling factors, and its superiority demonstrated comparing accuracy with that two other typical methods. results indicate PSCNN can effectively improve 32 times, an overall mean relative error three velocity components being less than 4.0%, reconstructed agrees well ground truth in distribution variation. Furthermore, reconstruct reasonable unseen conditions, further proving generalizability model capturing relationships wake. Overall, has significant applications obtaining snapshots fluid experiments.
Язык: Английский