基于代理模型和线性近似的快速气动热边界求解方法

Fast Method to Determine Thermal Boundary Based on Surrogate Model and Linear Approximation

采用本征正交分解(Proper Orthogonal Decomposition,POD)与Kriging代理模型相结合,建立快速求解气动热参数的降阶模型。引入一种基于换热系数的线性近似方法进行热边界传递,以减少因壁温变化引起的流场迭代。以三维翼面为例,对气动热的计算分析表明,利用POD-Kriging模型快速得到的翼面热流分布与计算流体力学(Computational Fluid Dynamics,CFD)方法得到的热流分布平均误差在7%以下,而且随着POD基模态个数的提高,误差能够明显降低。结果表明该方法具有较好的计算精度,结合线性近似方法处理热边界,在不降低气动热计算精度的前提下,能够大幅度减少计算耗时。

A reduced order frame combining the proper orthogonal decomposition(POD)with Kriging model was built for calculating the aerodynamic heat transfer quickly.A linear approximation method based on heat transfer coefficient is introduced to reduce the flow iterations due to the wall temperature change.Taking the 3D wing as an example,the result showed that the average error of the heat flux of the wing surface got from the POD-Kriging model is below 7%,while comparing with the CFD results.With the increase of the POD basic mode,the error could be greatly reduced,which shows that this method has good calculation accuracy.With the linear approximation method to deal with the thermal boundary,the computational time can be greatly reduced without reducing the accuracy of the aerodynamic thermal calculation.