论文部分内容阅读
大跨度屋盖结构脉动风荷载特性复杂,一般需要通过风洞试验确定。典型屋盖的风洞试验数据的积累为相似体型屋盖结构风荷载取值提供了依据。为拓展典型屋盖风洞试验数据的应用范围,基于广义回归神经网络,结合典型球面屋盖系列风洞试验建立了大跨度球面屋盖的风荷载预测模型。其中,风荷载由平均风压系数、脉动风压系数、偏度、峰度、3个自功率谱密度参数(包括高频段衰减斜率、无量纲谱峰值和无量纲峰值频率)以及互功率谱密度相干指数8个特征参数描述。通过交叉验证和试算确定了广义回归神经网络模型中的平滑因子取值。以安庆电厂球面网壳结构为例进行了风荷载预测,通过对比预测风荷载与风洞试验得到的风振分析结果,验证了预测方法的可行性。
Long-span roof structure pulsating wind load characteristics of complex, generally need to be determined by wind tunnel test. The accumulation of wind tunnel test data of a typical roof provides a basis for the wind load value of a similar body roof structure. In order to expand the application range of typical roof wind tunnel test data, a wind load prediction model of long-span spherical roof is established based on generalized regression neural network and wind tunnel test of typical spherical roof series. Wind load consists of average wind pressure coefficient, fluctuating wind pressure coefficient, skewness, kurtosis, three self-power spectral density parameters (including high frequency attenuation slope, non-dimensional peak and non-dimensional peak frequency) and cross power spectral density Coherence index eight characteristic parameters described. The value of the smoothing factor in the generalized regression neural network model is confirmed through cross-validation and trial calculation. Wind load forecasting is carried out by taking the spherical reticulated shell structure of Anqing Power Plant as an example. The wind-induced vibration analysis results obtained by wind load test and wind tunnel test are compared to verify the feasibility of the forecasting method.