03.2 - Experimental Aerodynamics
ADAPTIVE DESIGN OF EXPERIMENT BASED ON GAUSSIAN PROCESS MODELS WITH NONSTATIONARY KERNEL APPLIED TO WIND TUNNEL TESTING
U.H. Choi¹, J.H. Kim¹, Y.B. Lee, Agency for Defense Development, South Korea; J.M. Ahn¹; ¹KAIST, South Korea
Adaptive Design of Experiment (ADoE) based on Gaussian Process (GP) models with multiple kernel is applied to wind tunnel testing. Several GP kernels are analyzed in terms of the wind tunnel data regression. It is shown that accuracy, efficiency and robustness of the experiment is improved through the ADoE with several GP models even when the aerodynamic model property is rarely known.