中文摘要：

针对风压信号呈现的非高斯特性和传统极限估计方法的局限,首次提出基于保证率和相关性的极值估计方法——全概率迭代法进行冷却塔表面脉动风荷载极值分析,并和传统的峰值因子法及改进的Sadek-Simiu法计算结果进行对比验证。结果表明：全概率迭代法避开了对随机过程的高斯分布假定,相比传统的极值估计方法其结果更加真实可靠;表达风压极值中脉动分量的峰值因子数值沿着环向和子午向变化显著,如取为同一数值则偏于危险或过于保守;采用全概率迭代法得到的表面风压系数极值分布曲线与规范取值相比,迎风面和负压峰值区域极值偏小,背风区域极值偏大,且最小值对应角度相差约10°。

英文摘要：

Peak factor method and Sadek-Simiu method based on Gauss distribution are introduced in detail,the method of estimating extreme value based on reliability theory is potimized,and the total probability convergence method which is applied for single sample is proposed.The results of pressure test for ultra large-scale cooling tower calculated by these methods are discussed.The analysis results show that the total probability convergence method avoids the assumption of Gauss distribution and can gain better results for non-Gauss distribution.Because of the effect of turbulent flow and vortex shed its extreme value of wind responses in different areas of surface of tower is different,the same value to consider its pulse wind＇s effect should not be taken.In this paper,the distribution rule about sectional drag force coefficients along the tower height is analyzed based on probability correlation technique,then the Fourier serious fitting curves of wind pressure extreme value distributions along the circumferential direction for each section are also proposed.