中文摘要：

在利用风洞试验研究结构表面风压时,测点数量往往不能满足结构计算要求,为了得到未布置测点处的风压时程,需要进行插值预测。文章结合风洞试验同步采集到的风压时程数据,利用本征正交分解（POD）技术分析了定日镜表面风场分布特点;根据风压相关系数矩阵经POD分解后成为时间主坐标与空间本征向量这一特点,编制了Matlab程序,实现其与Surfer软件接口,选取距离反比法、三角形线性插值法和克里金法（Kriging）等不同插值方法对本征向量进行空间插值;研究了参与模态阶数对重构风压序列精度的影响,并计算了预测点风压时间序列。结果表明：随着参与重构的模态数量增加,重构与测量风压序列逐渐接近;在三种插值预测方法中三角形线性插值法和克里金法预测效果优于距离反比法。

英文摘要：

When wind tunnel test is used to research heliostat＇s surface wind pressure, number of measuring points usually cannot meet demand of computation, as requires interpolation prediction to obtain wind pressure time series of which where measuring points are not arranged. This article analyse heliostat＇s surface wind field distribution characteristics using technique of proper orthogonal decomposition （POD） method plus wind pressure time interval data synchronously collected in wind tunnel test. It＇s Matlab programmed according to point that POD decomposed wind pressure correlation matrix becomes time principal coordinates and spatial eigenvectors, and interfaced with Surfer software. Different interpolation methods like inverse distance method, triangulation with linear interpolation ＆ Kriging method are selected to spatially interpolate eigenvectors to study participant mode＇s exponents＇ influence on sequential accuracy of restructured wind pressure and calculate predicted points＇ wind pressure time series. The result indicates： restructured and surveyed wind pressure sequences gradually approaches to each other as restructuring-participant mode increases in number; among three interpolation prediction techniques, triangulation with linear interpolation and Kriging method are better than inverse distance method on forecast effect.