基于非线性变换的PCP-C模型及其在地下水动态分类中的应用

The Principal Component Projection-clustering Model Based on Nonlinear Transformation and Its Application in Groundwater Classification

为了克服目前地下水动态分类方法中存在的不能揭示分类指标空间到类型空间的非线性映射关系、方法复杂、计算量大等缺陷,可采用基于非线性变换的主成分投影(PCP)-聚类(C)模型,对地下水动态进行分类.方法首先对分类指标数据进行对数中心化变换,然后应用主成分投影法将变换后的多维指标向量映射到最优一维向量空间,并根据各样本指标在一维向量空间的投影值进行聚类分析,由此得到地下水动态分类结果.地下水动态分类结果表明,建议方法概念清晰,结构简单,计算简便,分类结果可信,是一种有效的地下水动态分类方法.

This paper is mainly to make dynamic classification of groundwater with the principal component projection - cluster model based on nonlinear transformation, aiming at the drawbacks such as being not able to reveal the non-linear mapping relationship be- tween space of classification indicators and the space of type, method complexity, having low computational efficiency exiting in the current dynamic classification of groundwater. In the application of the model, the indicators data of classification would be given a quasilinear transform through transformation of center of logarithmic, and a mapping from multi- dimensional vector space to the one-dimensional vector space using the method of principal component projection which can reduces the data noise. At last, the classification results can be found by cluster analysis according to the size of the projection value of the sample indicators in the one-dimensional vector space. The results of groundwater classification example show that the proposed method is an effective method of groundwater classification for dear concept, simple structure, simple computation and reliable classification result.