低矮房屋屋面脉动风压的预测

Prediction of fluctuating wind pressure on low building's roof

利用实测手段研究结构表面风压时,因测点布置数量有限,整个风场面貌不能充分反映,通常用插值法,但直接对时间序列插值难以实现。基于POD技术并借鉴地理统计学中"网格化"思想,对各阶空间本征向量进行处理,引入克里金法插值方法,编制与Surfer软件接口的Matlab程序,利用Surfer软件强大空间插值功能,对结构表面各阶本征向量进行空间插值。通过对台风发生时实测得到的低矮房屋屋盖风压时序的预测及误差分析验证该方法的可行性。研究表明克里金法可弥补三角线性插值法无法外插测点连线范围以外数值的不足,在风压变化梯度较大区域具有优于距离反比法的预测效果。该方法具有较强的通用性,可广泛应用于风洞试验和现场实测。

The number of taps is usually finite in structure＇s surface wind pressure inspection by measurement method, so the wind field＇s sufficient appearance is restricted. Interpolation is a regular choice, but it＇s hard to directly implement the interpolation of time series. Based on Proper Orthogonal Decomposition（POD） technique and referring to ＇grid＇ thought in geographical statistics, each phase of spatial latent vector was processed. The Kriging interpolation method was introduced to frame a Matlab program to interface with Surfer software whose powerful spatial interpolation capacity was utilized to spatially interpolate the structure＇s surface latent vectors of each phase. Data from measured wind pressure time series on a low-rise building in typhoon were used to verify the feasibility of the method. The results show that Kriging method can offset the deficiency of triangular linear interpolation which can not extrapolate the points out of the scope; moreover Kriging method is better than the inverse distance method as regards prediction effect when wind pressure varies sharply. The method is relatively universal, and can be widely applied in wind tunnel tests ＆ field measuring.