03.4 - Applied Aerodynamics
SUPERVISED DIMENSIONALITY-REDUCED MULTI-FIDELITY SURROGATE MODEL ASSISTED HIGH-DIMENSIONAL NATURAL LAMINAR FLOW OPTIMIZATION
Y. Gan¹, H. Zhao¹, S. Wang¹; ¹Northwestern Polytechnical University, China
A high-dimensional NLF optimization method based on a supervised partial least squares dimensionality-reduced multi-fidelity polynomial chaos–Kriging (PLS-DR-MFPCK) surrogate is proposed. Applied to an 88-dimensional transonic flying wing, it significantly improves efficiency and accuracy, achieving a 24.44-count (22.7637%) drag reduction while satisfying aerodynamic and geometric constraints.
DOI: 10.71945/icas2026_0502