.png){kind=icon}
.png){kind=icon}
Return
Transferable scaling function learning method for knowledge embedded aerodynamic database construction
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ast.2026.112097.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A transferable scaling function learning method is proposed for the efficient construction of aerodynamic databases. • Pre-training is performed using aerodynamic data from multiple source domain shapes, and aerodynamic scaling function expressions generalized to shapes are obtained through symbolic regression. • Aerodynamic model can be efficiently constructed by fine-tuning the scaling function parameters using only sparse aerodynamic samples of the new shape. • Compared to Kriging and DNN, the proposed method improves extrapolation prediction accuracy by 27–97 %.
Keywords:
scaling function
aerodynamic database
symbolic regression
transfer learning
extrapolation accuracy
Journal
IF:
5.8
Papers:
10.0K
Citations:
3.0W