摘要
为研究扁平钢箱梁风屏障的防风效果,在验证数值模型准确性的基础上,针对扁平钢箱梁上分离障条直线型风屏障,研究了风屏障对桥面风场分布的影响,采用神经网络模型建立了风屏障风速折系数的代理模型,考查了风屏障高度与透风率的影响,给出了风速折减系数随风屏障高度和透风率变化的等值线曲线。结果表明:当风屏障高度增加到一定程度后,再增加高度对风屏障防风效果的提高有限;风速折减系数随风屏障透风率的增加,在迎风侧车道和背风侧车道位置处呈现不同的形态;在风屏障透风率大于20%,高度小于3.5 m的情况下,上游车道的风速折减系数大于下游。风屏障风速折减系数神经网络模型可用于预测和评估风屏障的防风效果,研究结果为风屏障参数的选择提供参考。
In order to study the protective effect of the wind barrier on a flat steel box girder,based on the verification of the accuracy of the numerical model,the influence of the wind barrier on the wind field distribution of the bridge deck was studied.A neural network model was used as the surrogate model of the wind speed reduction factor of the wind barrier,the influence of height and porosity of the wind barrier on its wind protection effect was analyzed,and the contour of wind speed reduction factor was provided.The results show that:when the height of the wind barrier is increased to a certain extent,increasing the height of the wind barrier does not significantly improve the wind protection effect;with the increase of the porosity of the wind barrier,the wind speed reduction factor presents different curves at the positions on the windward side lane and the leeward side lane;when the porosity of the wind barrier is greater than 20% and the height is less than 3.5 m,the wind speed reduction factor on the upstream lane is greater than that on the downstream lane.The neural network model of wind speed reduction factor can be used to predict and evaluate the wind protection effect of the wind barrier,and the research results can be used as a reference for the selection of wind barrier parameters.
作者
胡博
向活跃
李永乐
HU Bo;XIANG Huoyue;LI Yongle(School of Civil Engineering,Southwest Jiaotong University,Chengdu 610031,China;Wind Engineering Key Laboratory of Sichuan Province,Southwest Jiaotong University,Chengdu 610031,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2022年第6期298-306,共9页
Journal of Vibration and Shock
基金
国家自然科学基金(51778544,51978589,51908077)。
关键词
风屏障
风速折系数
计算流体力学(CFD)模拟
神经网络代理模型
wind barrier
wind speed reduction factor
computational fluid dynamics(CFD)simulation
surrogate model of neural network