摘要
本文采用高斯过程回归方法,选取平坡屋面32个代表性测点(总测点的50%)的位置信息和风压系数为训练样本构建屋面风压系数的插值模型,分别得出0°,30°,60°和90°风向角下平均风压系数、脉动风压系数和峰值负压系数的插值结果。由插值结果和实测对比的误差发现,插值模型拥有较好的准确率,在0°和90°风向角下插值误差不超过26%(绝对值),危险风向角30°和60°下个别测点误差上升但最大不超过50%。由插值模型获取的风压系数云图能较好地补足试验测点难以顾及的边缘位置,使得云图更加平滑,屋面涡旋区域描述更清晰。高斯过程回归方法适用于风压系数插值研究,可有效减少风洞试验模型的测点布置数量,减轻试验繁琐度。本文研究成果可对风洞试验气动数据库提供一定支撑。
Gaussian process regression method is used to select the positional information and wind pressure coefficient of 32 representative measuring points(50% of the measuring points)of flat slope roof as the training model to construct the interpolation model of wind pressure coefficient of the building roof,and such interpolation results as average wind pressure coefficient,fluctuating wind pressure coefficient and negative peak pressure coefficient at 0°,30°,60° and 90° wind direction angles respectively are obtained. The error of the interpolation result and the actual measurement comparison shows that the interpolation model has a good accuracy. The interpolation error is not more than 26%(absolute value)at 0° and 90° wind direction angles,and the individual measuring point error at 30° and 60° dangerous wind angles increases but does not exceed 50%. The wind pressure coefficient cloud image obtained by the interpolation model can better complement the edge position that is difficult to be considered by the test points,making the cloud image smoother and the description of the roof vortex area clearer. The Gaussian process regression method is suitable for wind pressure coefficient interpolation,which can effectively reduce the number of measuring points in the wind tunnel test model,reduce the cumbersomeness of the test. The results could provide some supports for the wind tunnel test pneumatic database.
作者
严赫
吴健雄
许俊
Yan He;Wu Jianxiong;Xu Jun(Guangxi Key Laboratory of Geomechanics and Geotechnical Engineering,Guilin University of Technology,Guilin 541004,Guangxi;China Machinery International Engineering Design&Research Institute Co.,Ltd.,Changsha 410007,Hunan)
出处
《工程建设》
2020年第11期21-28,共8页
Engineering Construction
关键词
低矮建筑
高斯过程回归
风压系数插值
low rise building
Gaussian process regression
interpolation of wind pressure coefficients
