摘要
采用概率相关系数(PPCC)最大化原则和极大似然法,根据杭州湾中心岛屿玉盘山气象站的实测风速资料,分析不同风向的最优分布类型,探讨最优分布类型与概率曲线相关系数随形状参数变化规律之间的关系,并用风速风向联合分布推算设计基本风速。研究结果表明:在3种极值分布中的PPCC随形状参数递减的分布为最优,具体哪种分布的PPCC递减则依赖于具体风速样本,因此在以后的风速概率拟合中可以先由PPCC的规律确定分布类型,然后仅求解此分布的有效最优参数;16个风向中的最优分布除正西方向是极值Ⅱ型外,其他都是极值Ⅲ型;用联合分布推算的基本风速比传统的非联合分布的结果略低,传统的方法得到的基本风速偏保守,用联合分布推算基本风速更加科学。
Based on wind velocity datum of the Yupanshan weather-station in the center the Hangzhouwan bay, the maximization principle of probability plot correlation coefficient(PPCC) and the maximum verisimilitude method are adopted to analyze the optimal probability distribution type of wind velocity in all directions. The relation between the optimal distribution type and the variation trend of probability plot correlation coefficient is discussed with shape parameter of distribution. The joint distribution is used to estimate the designing basic wind velocity. Calculating results show that the type is always optimal whose PPCC decreases monotonically with shape parameter among three extreme distribution types. But what distribution types' PPCC accords the trend is dependent on wind velocity samples. So it can be put forward that, in probability fitting of wind velocity, the optimal distribution type could be assured according to the inherent relation between optimal type and PPCC and only the distribution's effective optimal parameters need to be solved. The Weibull distribution is optimal in 16 wind directions except for west direction in which the Frechet distribution is optimal. The basic wind velocity estimated by joint probability distribution is a little less than that by traditional non-joint distribution. Basic wind velocity by joint distribution is more scientific, and the traditional method is inclined to be conservative. 6 tabs, 4 figs 10 refs.
出处
《长安大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2008年第5期53-57,共5页
Journal of Chang’an University(Natural Science Edition)
基金
国家自然科学基金项目(50578014)
关键词
桥梁工程
基本风速
风速风向联合分布
极值分布
极大似然法
bridge engineering
basic wind speed
joint probability distribution of wind speed and wind direction
extreme distribution
probability plot correlation coefficient