Robust Aerodynamic Design Optimization via Variable Fidelity Surrogate Model

Abstract

In this paper, a robust design optimization framework is proposed with a variable fidelity Kriging model. By the use of the variable fidelity Kriging model approach, an accurate surrogate model can be constructed efficiently by the absolute values of a high-fidelity function as well as the trends obtained by low-fidelity function values. The highand low-fidelity levels can be defined by utilizing different physical models, computational meshes and so on. The robustness of a candidate design is efficiently evaluated by a Monte Carlo simulation which is executed on the variable fidelity Kriging model. The efficiencies of robust design optimization approaches are investigated in a 2D airfoil drag minimization problem. In this problem, free-stream Mach number as well as target lift coefficient are supposed as uncertain parameters. The mean and standard deviation of drag coefficient are simultaneously minimized to obtain non-dominated robust optimal designs. The developed robust design optimization approach via the variable fidelity Kriging model is shown to be useful for efficient search of robust airfoil designs.

Robust optimization of the aerodynamic design of a helicopter rotor blade based on multi-fidelity meta-models

“…6 Therefore, MFMBO strategies are adapted or extended to take into account uncertainty. For instance, Keane et al 35 and Yamazaki Wataru 36 used co-Kriging in the robust design to optimize the mean performance where the uncertainty propagation was applied on the MFMM. Ariyarit et al 37 proposed a multi-fidelity strategy based on a hybrid surrogate model and the EI refinement criterion.…”