基于主成分分析的河流洪水系统聚类法

进行河流洪水聚类的目的是根据洪水特征的相似程度划分洪水类别,研究同类洪水的规律性以及应对措施.但是,洪水特征选择过多往往会增加计算的复杂程度,同时特征之间的相关性也使得信息大量重叠,导致计算结果失真.为此,提出基于主成分分析的河流洪水系统聚类法.该法首先将所选的洪水特征综合成少数几个不相关的主成分,然后计算出每场洪水在各主成分上的得分值并将该值作为新的洪水特征值,最后根据这些新特征值进行洪水聚类.三门峡水库入库洪水聚类实例证明了该方法的可行性.

Flood clustering method based on principal component analysis

The aim of flood clustering is to classify the flood category by the similarity of flood characteristics,to study the law of similar floods and to find countermeasures,however the selection of too many flood characteristics always makes the calculation more complicated,and the correlation between the characteristics leads to a great amount of information overlapping and incorrect result of calculation.Therefore,the clustering method for river floods based on principal component analysis was proposed.In this method,the flood characteristics selected were classified into a few uncorrelated principal components,then the scores of each flood on each principal component were calculated and taken as the new characteristic values of flood.Finally flood clustering was performed according to these new characteristic values.The feasibility of the present method was verified by case study on inflow flood of Sanmenxia Reservoir.