一种数据挖掘方法-析取克立格法理论及其在品位估计中的应用

非线性现象广泛存在于许多自然科学研究之中。非线性空间信息统计学主要包括对数正态克立格、指示克立格、析取克立格、条件期望等理论方法,已在矿床储量估计中显示出其强大的生命力。其中析取克立格法(DK)介于线性地质统计学和条件期望之间的新方法,其前提假设Z(x)服从一、二元正态分布,并利用厄尔米特多项式展开进行估计,该方法更有效、准确、易实现,并能很好地估计Z(x)的任意函数,解决可回采储量估计等现实问题。该文主要研究了二元高斯分布法、正态变换和埃尔米特多项式的应用,并实现了析取克立格的算法,再现了它在真值空间变异性上的优势。

Data Mining--the Theory of Disjunctive Kriging and Its Application in Grade Estimate

Nonlinear phenomenon exists usually in the geoscience.Generally speaking,there are primarily logarithms gaussian kriging,indicator kriging,disjunctive kriging and condition expection about Nonlinear spatial information statistics.Disjunctive kriging(DK)is a new data mining method between Condition expection(CE)and Linear Geostatistics.It's more perfect and easier to be applied,which is also perfect for any function of Z(x)to solve problems in reality like estimate of recoverable reserve.The research contents of bivariate gaussian distribution,gaussian anamorphosis,varigram and struction analysis and Hermite polynomials's application.This study also carries out Disjunctive Kriging,manifests its advantage in spacial variability and proves the validity of the non-linear Geostatistics.