MACHINE LEARNING-BASED MODEL ORDER REDUCTION FOR THE GUST LOAD ANALYSIS

This paper presents an efficient nonlinear non-intrusive model order reduction (MOR) framework for the gust load analysis. The proposed method is based on artificial neural network (ANN), specifically a least-square hierarchical variational autoencoder (LSH-VAE). This approach will enable construction of nonlinear reduced-order model and allow accurate interpolation with regard to the parameters. The proposed method will be validated by applying for a high-altitude long-endurance (HALE) unmanned aerial vehicle (UAV). The accuracy and computational efficiency of the method will be compared against those by a full order model (FOM). It is found that the proposed method will construct accurate interpolated field with regard to the relevant parameters.