A new sample update strategy based on kringing
The study presented in this paper sets up a new mixing sample update strategy based on the Kriging model to address the issue of accuracy in complicated engineering optimization problems by adding points dynamically.First,a new optimized Latin Hypercube Sampling experiment design method is established to increase the homogeneity of the sample.Second,a sample update strategy based on local optimization and EI-based Kriging model is established.Three numerical samples are carried out to compare the results of traditional popular method EGO,and the results show that this method has increased the accuracy and the efficiency.The successful application of this method on RAE2822 Airfoil using less sample and reducing 14.3%of the drag has further proven the effectiveness of this method.
Latin Hypercube Sampling Kriging EGO Optimization
CHEN Yuanyuan BAI Junqiang WANG Dan
Northwest Polytechnical University,Xian 710072
国际会议
长沙
英文
1124-1131
2014-05-31（万方平台首次上网日期，不代表论文的发表时间）