Gaussian Process Regression Implemented as Surrogate Model to Aid Aerodynamic Design Process

Designing aircraft with optimal aerodynamic characteristics can be a difficult and timeconsuming task due to the large computational costs of current physics-based solvers and the large parameter space associated with aircraft design. The purpose of this study is to develop a robust surrogate model that can predict and represent aerodynamic loads of interest for aerodynamic design. In future studies this model will also serve as a framework to be coupled with multi – objective optimizers, to select an optimal geometry for different flight conditions. Our chosen surrogate model is a gaussian process regression model (GPR), and it will be coupled with high fidelity computational fluid dynamic commercial code, Star-CCM+. A low fidelity model will also be trained on airfoil sweeps done in XFLR5. The results will show the capabilities of GPR as a surrogate model and method to guide data collection to build a comprehensive analysis on flight loads and moments.