基于熵权的模糊聚类模型在山区融雪洪水产流类型中的分类应用

Flood-type Classification in Mountainous Catchments Using Fuzzy C-means Algorithm Classification Based on Entropy Weights

山区融雪洪水流量过程因洪水成因不同而异,但又表现出自身的规律性。将洪水产流类型分为若干类能够更好的掌握洪水机理和进行洪水预报。运用熵权确定影响因素的权重,并利用模糊C均值聚类模式识别分类方法得出洪水样本的自然分类。实验区位于新疆天山北坡山区地带的军塘湖流域。选取了12场洪水过程的观测资料,借助试验区架设的自动气象观测站的气象资料,积雪样方观测资料,将洪水的产流类型分为:超渗产流型,超渗蓄产流型,蓄满产流型。研究表明:(1)熵权能够有效地对影响洪水的各要素赋权,从而让模糊C均值聚类分析能够区分每个洪水样本的类型。(2)该方法能够准确、迅速的判断洪水产流类型,能有效地提高洪水预报精度。开展融雪洪水过程线分类研究,对于融雪洪水预报与预警有重要的理论和实际意义。

The mountainous area snowmelt flood flow varies with each causation, but shows regularity. Subdividing flood runoff generations into different types can know well the mechanism of the flood and make a correct flood prediction. This study uses entropy coefficient to deter- mine the weight of influence factors, and utilizes fuzzy C-means algorithm recognition classification method to get the natural classification of the flood samples. The experimental plot is situated in Juntang Lake drainage of the north slope of the Tianshan Mountains in Xinjiang. This paper chooses the observational data of 12 flood processes, with the aid of meteorological data from automatic weather station set in the experimental zone and observational data of snow samples, divided flood runoff types into over infiltration runoff generation type, super infiltration runoff generation type, saturation runoff generation type. Researches shows ： （1） entropy weights can effectively empower the influencing factors of the flood, and the fuzzy C-means algorithm analyse can distinguish between each flood sample types. （2） the method can accurately and quickly judge flood runoff generation type, can effectively increase the accuracy of flood prediction. Snow-melt flood hydrograph classified study has an important theoretical and practical significance in the prediction and alarming system of snow-melt flood.