基于广义Pareto分布的大坝预警指标拟定方法  

Warning Index Development of Dam Monitoring Based on Generalized Pareto Distribution

作　　者：张海龙[1] 马斌[1] 梁亚平[1] 

机构地区：[1]西京学院土木工程学院,陕西西安710123

出　　处：《水力发电》2017年第6期97-99,114,共4页Water Power

基　　金：陕西省教育厅科学研究项目(15JK2171)

摘　　要：将极值理论中的广义Pareto分布模型用于大坝预警指标拟定中是一个行之有效的方法,该方法通过设定某一阈值,以监测序列超过该阈值的实测值为分析样本,并认为该样本服从广义Pareto分布。由于阈值的选取关系到广义Pareto分布参数估计的精度,而用超限期望图法确定的阈值具有较大的主观性误差,因此,从定义出发,通过数学推导,建立了自动选取阈值的方法。通过实例分析表明,采用自动选取法确定阈值的广义Pareto分布拟定的大坝预警指标更加合理。It＇s an effective method to develop warning index of dam monitoring by using generalized Pareto distribution of extreme value theory. By setting a threshold value, the overrun samples are selected from monitoring data which exceed the threshold value and the samples are considered to obey generalized Pareto distribution. Since the accuracy of parameters of generalized Pareto distribution is affected by the selection of threshold value and the threshold value which is computed by exceeded expectation diagram method has a large subjective error, therefore, the auto select method of threshold value is established through mathematical derivation from the definition. The result of case study indicates that the warning index of dam monitoring which are developed by using generalized Pareto distribution are more reasonable.

关 键 词：大坝预警指标 广义PARETO分布 自动选取法 阈值 

分 类 号：TV698.1[水利工程—水利水电工程]