论文部分内容阅读
为了解决优化设计中计算效率与高可信度信息获取之间的矛盾,从高、低可信度模型的物理机理出发,基于Kriging模型和拉丁超立方设计选样方法构造两模型差值的代理模型;并利用该代理模型对低可信度模型进行修正,构成了具有高可信度的修正模型。与直接对高可信度模型构造的代理模型相比,修正模型不但分析精度更高,而且所需的构造样本更小。文中分别以翼型气动力分析、机翼气动力分析和无人机隐身特性分析为例,从不同维数、不同学科的角度验证了修正模型特性,并进行了机理分析。
In order to solve the contradiction between computational efficiency and high-confidence information acquisition in optimization design, this paper starts with the physical mechanism of high and low confidence model, and constructs the agent of two models difference based on Kriging model and Latin hypercube design sampling method Model; and using the proxy model to correct the low credibility model, constitutes a highly reliable correction model. Compared with the proxy model which is directly constructed for the high-confidence model, the revised model not only has higher analysis accuracy but also requires smaller sample construction. Taking aerodynamic analysis of airfoil, aerodynamic analysis of wing and stealth characteristics of UAV as examples, the characteristics of the modified model are validated from different dimensions and different disciplines, and the mechanism analysis is carried out.