03.4 - Applied Aerodynamics
CORRELATION MODELLING OF SURFACE INVISCID FLOW CHARACTERISTICS AND FRICTION DISTRIBUTION BASED ON MACHINE LEARNING
S. Zhao¹, W. Zhang¹; ¹NORTHWESTERN POLYTECHNICAL UNIVERSITY, China
This paper proposes a physics guided machine learning modelling approach for skin friction distributions with high generalizability. Compared to end-to-end deep learning modelling of distributed forces, the method achieves high modelling accuracy (drag error of about 2-3%) with a five-fold reduction in sample size, strong interpolation prediction capability and low dispersion of results.