基于Monte Carlo方法的极值风速估计误差分析及其应用  

Error analysis of extreme wind speed estimation based on Monte Carlo method and its application

作　　者：胡小浓 方根深[1,2] 赵林[1,2] 葛耀君[1,2] HU Xiaonong;FANG Genshen;ZHAO Lin;GE Yaojun(State Key Laboratory of Disaster Reduction in Civil Engineering,Tongji University,Shanghai 200092,China;Key Laboratory of Transport Industry of Bridge Wind Resistance Technologies,Tongji University,Shanghai 200092,China)

机构地区：[1]同济大学土木工程防灾国家重点实验室,上海200092 [2]同济大学桥梁结构抗风技术交通行业重点实验室,上海200092

出　　处：《建筑结构学报》2023年第3期175-184,共10页Journal of Building Structures

基　　金：国家自然科学基金项目(52108469,51978527);上海市浦江人才计划(20PJ1413600);桥梁结构抗风技术交通行业重点实验室(上海)自主研究课题青年基金项目(KLWRTBMC-07)。

摘　　要：采用有限实测风速样本估计重现期极值风速存在较大不确定性。为此,针对目前应用最为广泛的极大似然法估计广义极值分布,提出了基于Monte Carlo方法的极值风速估计误差分析算法,建立了依赖概型参数、样本容量和重现期的极值风速标准差广义数学表达式。结果表明:标准差随形状参数、尺度参数和重现期的增大而增大;随样本容量的增大而减小。结合我国4个气象站点(北京、喀什、福州、上海)实测风速数据,在进行数据地形修正、台风数据剔除的基础上,分析计算了各站点的重现期极值风速及其标准差,并与既有规范对比,当考虑抽样误差对极值风速预测的影响时,建议以一倍标准差作为量化标准更为合理。Due to limited measured wind speed samples, there is inevitably great uncertainty in the extreme wind speed estimation. Aiming at the most widely used method of fitting the generalized extreme value distribution by the maximum likelihood method, this paper proposed an extreme wind speed estimation error analysis algorithm based on the Monte Carlo method, established a generalized mathematical expression for the standard deviation of extreme wind speed depending on probability parameters, sample size and return period. It is found that the standard deviation increases with the shape and scale parameters and return period, and decreases with the increase of the sample size. Combined with the measured wind speed data of four meteorological stations in China(Beijing, Kashgar, Fuzhou, and Shanghai), based on the terrain correction of data and elimination of typhoon data, the extreme wind speed of a certain return period and its standard deviation of each station are analyzed and calculated, which is compared with the existing codes, and it is observed that it is more reasonable to use one standard deviation as the quantitative standard when considering the impact of sampling error on extreme wind speed prediction.

关 键 词：极值风速 误差分析 Monte Carlo方法 极大似然 广义极值分布 

分 类 号：TU311.3[建筑科学—结构工程]