基于KNN与ISOMAP的地球化学数据处理与应用研究
发布时间：2018-08-30 12:32
【摘要】：化探数据处理是勘查地球化学的一项重要内容,不同的数据处理方法直接影响着化探找矿的效果及效率。化探数据处理是应用数学方法和计算机技术,从化探原始数据中发现和提取有效信息,揭示化学元素与各种地质现象的内在联系,为地球化学找矿提供依据。如何科学有效地提取化探异常信息,并从大量异常中进行快速准确地筛选评价,以确定进一步找矿的靶区,则是决定化探找矿工作关键。地球化学元素含量值并不局限于正态分布或者对数正态分布,具有不连续性、突变性、非均匀性、多样性和随机性等特征,即非线性特征。在进行化探数据处理时,对于非线性的特征就要采用非线性的算法。本文以青海省大柴旦镇柴达木山南坡一带地区为例,采用传统统计方法、KNN算法、聚类分析、主成分分析、ISOMAP算法对研究区1:10000土壤地球化学测量数据进行分析处理,在了解研究区地质背景的基础上,圈定成矿远景区,为该区下一步地质勘探工作提供了工作靶区。从地球化学元素含量异常的评价与研究出发,利用了现代的数学方法和非线性分析的方法,挖掘化探数据中蕴含的成矿异常信息。通过与传统分析方法的对比表明,KNN分类算法对化探数据中的元素含量异常有很好的识别作用。利用主成分分析将Cu、Au、Zn、As、Sb、Pb六种元素分成了两组,并在此基础上圈定了两组元素的组合异常。利用ISOMAP同样将六种元素分为两组,圈定此时两组元素的组合异常。通过对比得到,用ISOMAP算法圈定的组合元素异常比主成分分析圈定的异常区域分布集中且形状规则。
[Abstract]:Geochemical data processing is an important part of exploration geochemistry. Different data processing methods directly affect the effect and efficiency of geochemical prospecting. Geochemical data processing is the application of mathematical methods and computer technology to discover and extract effective information from the original data of geochemical exploration, to reveal the inherent relationship between chemical elements and various geological phenomena, and to provide the basis for geochemical prospecting. How to extract geochemical anomaly information scientifically and effectively, and how to select and evaluate quickly and accurately from a large number of anomalies in order to determine the target area for further prospecting is the key to determine the geochemical prospecting work. The content of geochemical elements is not limited to normal distribution or logarithmic normal distribution. It has the characteristics of discontinuity, mutation, heterogeneity, diversity and randomness, that is, nonlinear characteristics. In the process of geochemical data processing, nonlinear algorithm should be used for nonlinear characteristics. Taking the south slope area of Qaidam Mountain in Dachaidan Town, Qinghai Province as an example, this paper uses the traditional statistical method, such as KNN algorithm, clustering analysis, principal component analysis (PCA) and ISOMAP algorithm, to analyze and process the geochemical data of 1: 10000 soil in the study area. On the basis of understanding the geological background of the study area, the metallogenic area is delineated, which provides a working target area for the further geological exploration in this area. Based on the evaluation and study of geochemical element anomaly, the information of metallogenic anomaly contained in geochemical exploration data is excavated by using modern mathematical method and nonlinear analysis method. The comparison with the traditional analysis method shows that the KNN classification algorithm has a good effect on identifying the anomaly of element content in geochemical data. The six elements of Cu,Au,Zn,As,Sb,Pb are divided into two groups by principal component analysis, and the combined anomalies of the two groups of elements are delineated on this basis. The six elements are also divided into two groups by ISOMAP, and the combined anomalies of the two groups are delineated. By comparison, it is found that the combined element anomalies delineated by ISOMAP algorithm are more concentrated and regular in shape than those delineated by principal component analysis (PCA).
【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
本文编号：2213077
[Abstract]:Geochemical data processing is an important part of exploration geochemistry. Different data processing methods directly affect the effect and efficiency of geochemical prospecting. Geochemical data processing is the application of mathematical methods and computer technology to discover and extract effective information from the original data of geochemical exploration, to reveal the inherent relationship between chemical elements and various geological phenomena, and to provide the basis for geochemical prospecting. How to extract geochemical anomaly information scientifically and effectively, and how to select and evaluate quickly and accurately from a large number of anomalies in order to determine the target area for further prospecting is the key to determine the geochemical prospecting work. The content of geochemical elements is not limited to normal distribution or logarithmic normal distribution. It has the characteristics of discontinuity, mutation, heterogeneity, diversity and randomness, that is, nonlinear characteristics. In the process of geochemical data processing, nonlinear algorithm should be used for nonlinear characteristics. Taking the south slope area of Qaidam Mountain in Dachaidan Town, Qinghai Province as an example, this paper uses the traditional statistical method, such as KNN algorithm, clustering analysis, principal component analysis (PCA) and ISOMAP algorithm, to analyze and process the geochemical data of 1: 10000 soil in the study area. On the basis of understanding the geological background of the study area, the metallogenic area is delineated, which provides a working target area for the further geological exploration in this area. Based on the evaluation and study of geochemical element anomaly, the information of metallogenic anomaly contained in geochemical exploration data is excavated by using modern mathematical method and nonlinear analysis method. The comparison with the traditional analysis method shows that the KNN classification algorithm has a good effect on identifying the anomaly of element content in geochemical data. The six elements of Cu,Au,Zn,As,Sb,Pb are divided into two groups by principal component analysis, and the combined anomalies of the two groups of elements are delineated on this basis. The six elements are also divided into two groups by ISOMAP, and the combined anomalies of the two groups are delineated. By comparison, it is found that the combined element anomalies delineated by ISOMAP algorithm are more concentrated and regular in shape than those delineated by principal component analysis (PCA).
【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】
相关期刊论文 前2条
1 吴敏金;分形信息论及其应用[J];华东师范大学学报(自然科学版);1996年01期
2 周靖;刘晋胜;;一种采用类相关度优化距离的KNN算法[J];微计算机应用;2010年11期
，本文编号：2213077
本文链接：https://www.wllwen.com/kejilunwen/diqiudizhi/2213077.html
