32th Congress of the International Council of the Aeronautical Sciences

03.1 - Aerodynamics – CFD Methods and Validation

RESEARCH OF TURBULENCE MODELING BASED ON MACHINE LEARNING

Y.-H. Yin¹, Y.-F. Zhang¹; ¹Tsinghua University, China

The multi-layer perceptron neural network algorithm is adopted to train the mapping model from the average characteristic of the flow field to the Reynolds stress. The prediction of the eddy viscosity and Reynolds stress components can both have correct correction for the average field, and the smoothness problem needs further concern.


View Paper