基于核Fisher判别分析的地球化学异常识别

发布时间：2018-04-09 16:05

本文选题：核Fisher判别　切入点：地球化学异常　出处：《成都理工大学》2015年硕士论文

【摘要】：在地球化学矿产勘查研究中,进行地球化学异常评价重要环节是识别地球化学异常是非矿致异常或是矿致异常,进而圈定合理的找矿远景区。对地球化学异常区域的有效圈定不但有利于人们寻找矿区,而且可以缩减物力,节省财力,减少人力资源。因此本论文研究的重点是对地化地球化学异常进行识别。在数据处理过程中由于地球化学的大部分元素含量指标通常表现为互相影响和相互关联的形式,在对地球化学异常进行评价的时候,不能仅仅考虑少数元素含量指标,应尽可能多的考虑元素综合指标,这有助于提高异常识别的准确性。因此,地球化学元素综合异常或组合异常是判断地球化学矿致异常的有效方法。Fisher判别分析是一种有效的分类方法,它通过选择最佳的投影向量将数据进行投影变换,从而达到将数据进行分类的目的。这一思想可以为矿致异常的圈定提供数据支撑。然而,由于地质系统的复杂性,导致了地球化学元素信息表现出模糊性与非线性。在Fisher判别分析基础上定义的线性判别函数就不能很好的表示这种关系。因此,引入核函数更适合对地球化学数据进行判别。核方法指的是将线性不可分的输入空间数据非线性映射到高维特征空间,这样不可分的空间数据变为可分数据,从而可以采取线性的方法在特征空间上进行数据的特征提取。因此,核Fisher方法是比Fisher判别分析方法更有效的非线性分类方法。本文研究工作是利用基于核函数的Fisher判别分析对地球化学异常进行识别。基于核函数的Fisher判别分析对异常进行分类判别的原理是指借助“核技巧”将输入数据空间隐式地变换到一个非线性的特征空间,这样在变换的空间上利用线性的Fisher判别分析对数据进行异常识别。通过对鄂东南地区的1:20万水系地球化学数据的研究,表明基于核函数的Fisher判别分析在对地球化学异常识别方面效果显著。
[Abstract]:In the research of geochemical mineral exploration, the important link of geochemical anomaly evaluation is to identify the geochemical anomaly as non-mineral-induced anomaly or ore-induced anomaly, and then to delineate a reasonable prospecting area.The effective delineation of geochemical anomaly area is not only helpful for people to search for mining area, but also can reduce material resources, save financial resources and reduce human resources.Therefore, the focus of this paper is to identify geochemical anomalies.In the process of data processing, because most of the element content indexes of geochemistry usually show the form of mutual influence and correlation, when evaluating geochemical anomalies, we can not only consider a few element content indexes.As many elements as possible should be considered, which is helpful to improve the accuracy of anomaly recognition.Fisher discriminant analysis (Fisher discriminant analysis) is an effective classification method. It can transform the data by selecting the best projection vector.In order to achieve the purpose of data classification.This idea can provide data support for the delineation of ore-induced anomalies.However, because of the complexity of geological system, geochemical element information shows fuzziness and nonlinearity.The linear discriminant function defined on the basis of Fisher discriminant analysis can not express this relationship well.Therefore, the introduction of kernel function is more suitable to judge geochemical data.Kernel method refers to the nonlinear mapping of input space data from linear inseparability to high dimensional feature space, so that the inseparable spatial data can be transformed into separable data, so that the feature extraction of data can be carried out in the feature space by linear method.Therefore, kernel Fisher method is more effective than Fisher discriminant analysis in nonlinear classification.In this paper, Fisher discriminant analysis based on kernel function is used to identify geochemical anomalies.The principle of classifying anomalies by Fisher discriminant analysis based on kernel function is to implicitly transform the input data space into a nonlinear feature space by means of "kernel technique".In this way, the linear Fisher discriminant analysis is used to identify the anomaly of the data in the space of transformation.Based on the geochemical data of 1: 200 000 drainage system in southeast Hubei, it is shown that the Fisher discriminant analysis based on kernel function is effective in identifying geochemical anomalies.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P632

【参考文献】