基于主成分分析的FCM法在泥石流分类中的应用

选取流域面积、最大高差等12个传统泥石流沟评价因素和圆度、平均长度等4个沟谷形态统计要素对泥石流性质进行评价；利用主成分分析法对分析因素进行降维，将16个分析因素降维形成线性无关的6个主成分，大大减小了分析的复杂性，并且证明了考虑沟谷形态统计要素的必要性；并利用FCM法(模糊C-均值聚类法)对研究区域27条泥石流沟进行分类，把泥石流沟分为3类。结合分类结果及现场调查，证明分类结果与现场的实际情况吻合较好，并针对不同的分类提出泥石流防护措施的建议。

Application of FCM Based on Principal Components Analysis in Debris Flow Classification

Twelve conventional evaluation-factors such as watershed area, maximum altitude difference and four statistical factors of debris flow gully form such as circular degree, mean length are chosen to evaluate the debris flow characteristics. Principal component analysis is used to reduce the dimension of the evaluation-factor.Sixteen factors is classified into six uncorrelated principal components, which reduces the complexity in analysis process. The factors of debris flow gully form are proved important. FCM( Fuzzy C-means algorithm) is proposed to sort twenty seven debris flow field investigated into three grades.The result preferably coincides with practical situation of debris flow gullies. Guard measures are proposed according to the result of cluster.