信息矩阵方法在山洪灾害风险预警中的应用

发布时间：2018-03-27 18:11

  本文选题：信息矩阵方法　切入点：信息分配　出处：《计算机工程与应用》2017年14期

【摘要】：山洪灾害风险预警指标的确定是防洪减灾工作中的关键技术问题,但由于山洪实测资料序列较短且山洪灾害样本量较小,难以采用传统的统计模型建立山洪灾害风险预警指标。任何统计模型其本质是建立输入-输出关系,在数学上可以用函数关系表达,信息矩阵方法能够从小样本获得信息并建立这种函数关系。利用淠河流域历史降水与流量极值资料,将2003—2012年17次典型洪水过程作为学习样本,构建流域前期土壤含水量饱和度、5个时间尺度(1,3,6,12,24 h)降雨量以及4个重现期(5、≥5、≥20、≥50 a)洪峰流量的模糊关系矩阵,进而建立山洪灾害风险预警不同等级组合指标。从2003—2012年洪水资料中选取未参与统计的40组独立样本,对不同时间尺度山洪灾害风险预警组合指标进行应用检验,总体预警合格率达到70%,表明该方法应用于山洪灾害风险预警是可行的。
[Abstract]:The determination of early warning index of mountain torrent disaster risk is a key technical problem in flood control and disaster reduction work. However, because of the short sequence of measured data and the small sample size of mountain flood disaster, It is difficult to use traditional statistical model to establish early warning index of mountain torrent disaster risk. The essence of any statistical model is to establish input-output relation, which can be expressed mathematically by function relation. The information matrix method can obtain information from small samples and establish this functional relationship. Using the historical precipitation and discharge extremum data of Pi River Basin, 17 typical flood processes from 2003 to 2012 are used as learning samples. The fuzzy relation matrix of soil water saturation in the early stage of watershed, rainfall in 5 time scales (1 ~ 3 ~ 3 ~ 6 ~ 6 ~ (-1) ~ (-1) ~ (-1)) and runoff of Hong Feng in 4 recurrence periods (n = 5, 鈮，

本文编号：1672659