代理模型由于可以有效地缩减学科分析时间，被广泛应用于优化领域。而构建高精度代理模型则取决于样本点在设计空间中的分布。为了建立拟合效率高的近似模型，在已有Kriging代理模型基础上，提出一种基于垂距和最大化点均方差取样(Integrated mean square error, IMSE)的多点取样算法，以保证预测精度的同时减少样本数量。该方法将垂距作为判定设计变量取值的标准，进行数据点的初步筛选。选取高斯函数作为设计点之间的相关函数，并且在边缘附近进一步修正。针对实际算例，将该取样算法与多点加点准则比较，结果表明使用的方法在保证全局精度的基础上，采用较少的采样点构建代理模型，具有较高的局部近似精度。

Since the surrogate model can reduce disciplinary analysis time effectively, it is widely used in optimization. The accuracy and the calculation of surrogate model depend on the sampling points in the design space. In order to establish approximation model fitting the data well, an adaptive sampling algorithm based on Kriging model is put forward. The algorithm is based on vertical distance and integrated mean square error(IMSE)criterion to ensure prediction accuracy while reducing the number of samples. The vertical distance is adopted as the standard to decide the design variables and Gauss function as correlation function for design points Then the sampling region of experiment design is updated near the edge of the contour. Taking a practical example, the proposed algorithm is compared with the multi-point sampling criterion, the results show that the agent model built by the proposed method can effectively search both the local and global optimum using less sampling points.