咨询与建议

看过本文的还看了

相关文献

该作者的其他文献

文献详情 >An improved adaptive sampling ... 收藏

An improved adaptive sampling and experiment design method for aerodynamic optimization

An improved adaptive sampling and experiment design method for aerodynamic optimization

作     者:Huang Jiangtao Gao Zhenghong Zhou Zhu Zhao Ke 

作者机构:Computational Aerodynamics Institute China Aerodynamics Research and Development Center National Key Laboratory of Aerodynamic Design and Research Northwestern Polytechnical University 

出 版 物:《Chinese Journal of Aeronautics》 (中国航空学报(英文版))

年 卷 期:2015年第28卷第5期

页      面:1391-1399页

核心收录:

学科分类:080103[工学-流体力学] 08[工学] 080104[工学-工程力学] 0801[工学-力学(可授工学、理学学位)] 

基  金:co-supported by the National Natural Science Foundation of China (Nos. 11402288 and 11372254) 

主  题:Aerodynamic optimization Crowdness enhance function RBF model RCE adaptive sampfing RMSE feedback 

摘      要:Experiment design method is a key to construct a highly reliable surrogate model for numerical optimization in large-scale project. Within the method, the experimental design criterion directly affects the accuracy of the surrogate model and the optimization efficient. According to the shortcomings of the traditional experimental design, an improved adaptive sampling method is proposed in this paper. The surrogate model is firstly constructed by basic sparse samples. Then the supplementary sampling position is detected according to the specified criteria, which introduces the energy function and curvature sampling criteria based on radial basis function (RBF) network. Sampling detection criteria considers both the uniformity of sample distribution and the description of hypersurface curvature so as to significantly improve the prediction accuracy of the surrogate model with much less samples. For the surrogate model constructed with sparse samples, the sample uniformity is an important factor to the interpolation accuracy in the initial stage of adaptive sam- pling and surrogate model training. Along with the improvement of uniformity, the curvature description of objective function surface gradually becomes more important. In consideration of these issues, crowdness enhance function and root mean square error (RMSE) feedback function are introduced in C criterion expression. Thus, a new sampling method called RMSE and crowd- ness enhance (RCE) adaptive sampling is established. The validity of RCE adaptive sampling method is studied through typical test function firstly and then the airfoil/wing aerodynamic opti- mization design problem, which has high-dimensional design space. The results show that RCE adaptive sampling method not only reduces the requirement for the number of samples, but also effectively improves the prediction accuracy of the surrogate model, which has a broad prospects for applications.

读者评论 与其他读者分享你的观点