全局性气动外形优化中的流场加速求解新方法研究

项目名称： 全局性气动外形优化中的流场加速求解新方法研究

项目编号： No.11502211

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 数理科学和化学

项目作者： 邱亚松

作者单位： 西北工业大学

项目金额： 18万元

中文摘要： 全局性气动外形优化系统需要求解大量流场，由此产生了巨大的计算开销。代理模型方法能够大幅减少优化系统调用流场求解器的次数，但当设计变量较多时，为了获取构建代理模型所需的样本也需要求解不少的流场，计算开销仍然很大。针对以上问题，本项目提出了通过预测初始流场以加快流场求解速度的思想，即在求解一个流场之前先快速预测一个较为接近最终收敛解的初始流场，然后在此初始流场的基础上进行流场求解。按照这种思想，本项目将开展基于几何相似的初始流场预测方法及基于本征正交分解与代理模型的初始流场预测方法研究。在此基础上进一步开展两种流场预测方法的搭配使用以及预测初始流场方法加速不同CFD求解方法的效果研究。本项目旨在加速全局性气动外形优化系统中每次流场求解的速度，在代理模型技术的基础上进一步缩短流场求解气动优化中所占用的时间，增强全局性气动外形优化系统的实用性。

中文关键词： 全局性气动外形优化；流场求解；代理模型；本征正交分解；几何相似

英文摘要： In global aerodynamic shape optimization systems, huge amount of computation cost is needed for solving flow field. Although surrogate models could reduce computation cost dramatically, but in cases with large number of design parameters, the computation cost for calculating samples to construct surrogate model will still be unacceptable. To deal with this problem, a novel flow field solving accelerating method, based on initial flow field prediction, is proposed in this project. The essence of this method is to accelerate the convergence of flow field iteration by starting from an initial flow field, which is predicted to resemble the converged flow field. To realize this method, this project will research two initial flow field prediction methods, the first one is based on geometric similarity and the other one is based on proper orthogonal decomposition and surrogate model. Furthermore, this project will research the accelerating effects using different CFD methods. This research will reduce the computation cost of solving flow field in aerodynamic shape optimization, decrease the time expense of surrogate model based optimization methods and enhance the usefulness of global aerodynamic shape optimization system.

英文关键词： Global aerodynamic shape optimization;Flow field solving;Surrogate model;Proper orthogonal decomposition;Geometric similarity