POD方法在双坡屋盖风压场预测中的应用

应用本征正交分解(POD)技术对双坡屋盖的随机风压场进行了预测。结合风洞试验同步测量的风压数据,采用了两种方案预测屋盖上未布测点位置的风压系数时间序列。预测出数据的统计特性(均值、均方根值及极值)和功率谱等与实测数据进行了比较,发现所预测风压系数时间序列的均值和均方根值精度较好。虽然预测的风压极值大部分被低估,但是预测数据的功率谱在低频段与实测数据相吻合。因此将预测的数据应用于结构的风致响应计算是可行的。

APPLICATION OF POD TO PREDICTION OF WIND PRESSURE FIELDS OF PITCHED ROOF

The random wind pressure field of a pitched roof is predicted in terms of the proper orthogonal decomposition (POD) technique. With the wind pressure data which are measured simultaneously in a wind tunnel test, two schemes are adopted to predict wind pressure time series on the locations of a roof where the pressure taps are not distributed. The statistics, such as the mean value, root mean square value, peak value, and auto-spectra of the predicted data have been compared with those of the measured data. It is found that the mean value and r.m.s value of the predicted wind pressure time series are reasonably accurate. Most of the predicted wind pressure peak values are somewhat underestimated, however, the auto-spectra of predicted data are consistent with those of the measured data at the low frequency range. This leads to the feasibility of employing the predicted results in the wind-induced dynamic response analysis.