摘要
分析对比了传统模式识别方法（贝叶斯，费舍尔方法）与人工神经网络方法的应用效果，根据已有的地球物理资料，结合金刚石原生矿的成因模式，从金伯利岩产出的地质环境、指示矿物、地球物理和地球化学异常等方面提取了２０种用于识别金伯利岩物化探异常的特征，运用人工神经网络方法对鲁西地区１２×１０４ｋｍ２的物化探异常进行识别和筛选，从中选出了１０个一二类远景异常点供查证。
The application effect of artificial neutral network is analysed and compared with that of traditional pattern recognition methods(Bayes method and Fisher method). According to the acquired geophysical data and the genetic model of diamond primary ore, 20 kinds of features, which have been taken out of such respects of kimberlite as geologic setting, index mineral, geophysical and geochemical anomalies etc. are applied to identify the geophysical and geochemical anomalies of kimberlite. And 12000 square kilometres geophysical and geochemical anomalies in Luxi area are identified and sieved by the method of artificial neutral network(BP,ART), from which 10 class one or class two prospective anomalous points are selected for examination.
出处
《现代地质》
CAS
CSCD
北大核心
1998年第4期598-602,共5页
Geoscience
基金
地质矿产部地质调查局定向研究项目
关键词
人工神经网络
金伯利岩
地球化学勘探
artificial neutral network, kimberlite, geophysical and geochemical anomalies, Luxi area
