基于代理模型的气动外形平面参数多目标匹配设计

将Kriging代理模型和Pareto遗传算法引入气动外形平面参数匹配设计中,提出一种基于代理模型的多目标平面参数匹配设计方法。将拉丁超立方试验设计用于平面参数筛选,确定出参数匹配方案库;基于方案库的计算流体力学(CFD)分析结果构建Kriging气动代理模型;将Kriging模型替代CFD分析,用于气体布局参数匹配优化设计,提高了设计效率并保证了可信度;通过Pareto遗传算法优化解决多点设计要求下气动布局参数匹配问题,一次性给出参数匹配方案的最优解集,从Pareto前沿中根据设计偏向选择气动布局最佳匹配方案。以典型的双后掠布局平面参数多点匹配优化设计问题作为算例,研究结果表明:Kriging气动代理模型与实际CFD分析结果的误差均小于5%,满足精度要求;根据不同设计偏向,选择了3种参数匹配Pareto优化方案,与原样本方案相比超声速阻力减小6.0%~12.8%,跨声速升阻比增加0.01%~3.40%,证明了匹配设计方法的有效性;通过试验设计的Pareto分析与主、交互效应分析,获得了气动布局平面参数对气动性能影响的定量关系,能够为参数匹配设计提供依据。所提出的平面参数匹配设计方法可应用于其他常规与非常规气动布局型式。

Multi-object Aerodynamic Configuration Parameter Design Using Kriging Approximation

By applying the Kriging model and the Pareto optimization method, a multi object planform parameter design method based on approximation is proposed. The design of experiment is used to choose the configuration parameters and select configuration concept samples. The configuration samples are analyzed using the CFD, and the results are used to create the Kriging approximation model. Application of the Kriging model to parameter optimization can improve design efficiency and guarantee its precision. The Pareto multi object genetic algorithm is used for parameter optimization, and the planform optimum results can be chosen in the Pareto front. A cranked sweep configuration is selected as a design case. The design results indicate that Kriging approximation model can satisfy the requirement of precision with errors of less than 5%. Three planform optimums are chosen by different preferences in the Pareto front. Through optimization, the supersonic drag coefficient can be reduced by about 6.0%~12.8%, while the transonic lift to drag ratio can be increased by about 0.01%~3.40%. By Pareto of design of experiment, main effect and inter effect analysis, the quantitative relationship between the parameters and aerodynamic characteristics can be achieved. This method can be applied to other aerodynamic configurations including unconventional configurations.