01.1 - Aircraft Design and Integrated System (Basics and Theory)
LATENT SPACE CORRELATION FOR INTERPRETABLE AIRFOIL PARAMETERIZATION USING VARIATIONAL AUTOENCODERS
K. Swannet¹, C. Varriale¹, N.A.K. Doan¹; ¹Delft University of Technology, Netherlands
This study utilizes Variational Autoencoders (VAEs) to enhance airfoil design by improving latent space interpretability of geometric and aerodynamic features. It develops a comprehensive parameterization model, tackling dataset inconsistencies and broadening property extraction, aiming to offer an efficient, user-friendly solution for diverse engineering applications.