摘要
为满足勘查地球化学对化探数据处理越来越精细的要求,把人工神经网络和统计分析结合起来形成地球化学场精细结构解析方案。该方案包括对地球化学样品的分类和对各类样品性质的详细研究,目的在于提供便捷的数学和计算机软件工具,以提取化探数据中的深层信息。可以用于:①研究各组样品中元素含量水平和组合特征以解释可能的矿种;②研究元素组合特征与指示元素的关联和区别以揭示可能的成矿作用过程;③研究异常样品的空间组合属性以揭示地球化学场的精细结构;④研究异常样品之间的差异性以缩小找矿靶区;⑤综合分析以确定剥蚀深度和找矿前景。该方案也适用于环境地球化学及矿产资源预测研究。
In order to meet the demand for finer processing of geochemical data, an approach to the analysis of the fine structure of the geochemical field is designed, which combines the Kohonen artificial neural network with statistical algorithms. This approach includes the classification of geochemical samples and detailed studies of the properties of various kinds of sample. The purpose is to provide an easy and rapid mathematic and computer software tool so as to extract the deep-level information in chemical data. It may be used in accomplishing the following tasks: (1) studying the contents of elements in various groups of samples and assemblage features to assert possible valuable minerals; (2) studying the relation and difference between the features of element assemblages and indicator elements to reveal possible ore-forming processes; (3) studying the spatial distribution of anomalous samples to reveal the fine structure of the geochemical field; (4) studying the differences of the anomalous samples to reduce prospecting target areas; and (5) carrying out an integrated analysis to determine the denudation depth and ore prospects. This approach is also applicable to the environmental geochemical study and prediction of mineral resources.
出处
《地质通报》
CAS
CSCD
北大核心
2004年第2期147-153,共7页
Geological Bulletin of China
基金
863项目(2001AA135120)
973项目(G1999045708)
自然科学基金项目(49973015)
法国国家科研中心国际合作项目(PICS2039)共同资助
关键词
信息处理
知识挖掘
地球化学
人工神经网络
非线性
区域化变量
information processing
knowledge excavation
geochemistry
artificial neural network
non-linearity
regionalized variable
