Application of Adaptive Kriging Approximation Model in Two Dimensional Diffuser Aerodynamic Optimization Design

Zhimin Du
2011-01-01
Abstract:An adaptive global optimization method was developed coupled with uniform experimental design,CFD analysis,Kriging approximation model and niching micro genetic algorithm.In the optimization procedure,EI function was introduced to identify the next sampled point by considering the prediction and mean squared error of Kriging model to decrease the risk of trapping into the local optimum when the optimal strategy was used.The proposed method and niching micro genetic algorithm have been applied to the diffuser optimization design respectively,average static pressure recovery coefficient is selected as objective function and Nurbs curve is used to parameterize the geometric model.8.5% improvement of average static pressure recovery coefficient is obtained and the result shows that the method is more effective than stochastic optimization algorithm.
What problem does this paper attempt to address?