Optimization of a Hypersonic Inlet Based on Surrogate Models

Xianyu WU,Shibin LUO,Xiaoqian CHEN,Zhenguo WANG

DOI: https://doi.org/10.3969/j.issn.1673-9728.2008.01.045

2008-01-01

Abstract:An optimization strategy based on surrogate models was applied to hypersonic inlet optimization.The sample data points were selected by Latin hypercube design of experiment technique,and the sample database used to create surrogate models were constructed by inlet flow field calculation with computational fluid dynamics(CFD) method based on 2D viscous Navier-Strokes equation.Four types of surrogate models including polynomial response surface,Kriging,BP neural network(BPNN) and radial bases function neural network(RBFNN) were adopted and compared.For hypersonic inlet optimization,effects of the surrogate models on optimization were different,and the former two were more available.The update of surrogate models enhanced the potential of obtaining optimal inlet.The flow capture and compression performance of optimal inlets were improved,and total pressure recover increased except the BPNN optimization case,while drag coefficient also increased.