VAE-Based Surrogate Modeling of Fixed Pitch Rotor Airloads

Achieving noise reduction in rotorcraft requires an analysis of various design parameters and flight conditions. However, high-fidelity methods are computationally expensive. To overcome this limitation, reduced order model (ROM)-based surrogate models have been applied to aerodynamics and aeroacoustics prediction. This study proposes a ROM-based surrogate model employing a variational autoencoder (VAE) to predict rotor aerodynamic loads and associated noise. Train and test datasets were generated using reformulated vortex particle method across a wide range of flight conditions. The proposed framework was applied to a single rotor, and its performance was evaluated qualitatively and quantitively in comparison with proper orthogonal decomposition (POD)-based surrogate model. The results show that VAE-based model consistently outperformed the POD model in noise prediction. These results demonstrate that the proposed framework enables accurate rotor noise prediction under various flight conditions and provides a promising approach for low-noise rotorcraft design and operational planning.