STUDY ON THE IMPROVED KRIGING-BASED OPTIMIZATION DESIGN METHOD

摘要

The improved Kriging-model-basedrnoptimization design method which is combinedrnwith design of experiment of Latin hypercubernsampling, Kriging model and geneticrnoptimization algorithm is developed in thisrnpaper. By simultaneously adding the samplernpoint with maximum EI (Expected Improvement)rnand the optimal point from optimization of therninitial samples, a new Kriging model of higherrnaccuracy is formed gradually, with abovernmeasure the optimization of given objective canrnbe realized.rnThe fitting accuracy of Kriging model based onrnEI method is investigated and validated throughrnthe tests of a one- dimensional function and anrnaerodynamic problem, which show that therndeveloped Kriging model can be effectivelyrnused in objective evaluations in optimizationrnproblems.rnFor construction of the improved Kriging modelrnof the aerodynamic problems, the aerodynamicrnperformances of sample points are evaluatedrnusing a Reynolds Averaged Navier-Stokesrn(RANS) Solver. A drag reduction optimizationrndesign of RAE2822 airfoil is carried out forrnexamining the validity and efficiency of presentrnmethod. The drag coefficient of RAE2822 airfoilrnis reduced by 33.6%. It shows that this methodrncan gradually improve the fitting accuracy ofrnKriging model, finally achieve the greatrnimprovement of aerodynamic performance forrnthe airfoil.