自组织特征映射网络在复杂油气层识别图版研制中的应用  

Applications of Self-organizing Feature Map Networks to Chart Development of Complex Hydrocarbon Formation Recognition

作　　者：文环明[1] 肖慈珣[1] 李薇[2] 智刚[2] 

机构地区：[1]成都理工大学信息工程学院 [2]河南油田研究院

出　　处：《测井技术》2003年第1期47-50,89,共4页Well Logging Technology

摘　　要：自组织特征映射网络是一种非线性方法 ,主要用于无先验知识样本的聚类分析。首先利用该方法对有先验知识的学习样本进行聚类分析 ,并将学习样本映射到二维拓朴空间 (平面图 ) ,然后根据不同类型学习样本在拓朴空间的分布区域确定相应类型的判别界限。在复杂油气层识别图版的研制中 ,该方法比交会图法具有明显的优势 ,并在实际应用中取得显著效果。The self-organizing feature map network is a kind of nonlinear method that is mainly applied to the clustering analyses of samples without prior knowledge.Here is a different methodology from the old one. At first, clustering analyses for the learning samples with prior knowledge are made through the self-organizing feature map network. And then the learning samples are projected to a 2D topological space (planar map). Finally, the recognition standards are defined by means of the different distribution area responding to the type of samples in the 2D topological space.The method has more advantages than the crossplot in the chart development of complex hydrocarbon formation recognition. The good application has been available in the field's work.

关 键 词：自组织特征映射网络 复杂油气层 识别 图版 测井解释 

分 类 号：P618.13[天文地球—矿床学] P631.84[天文地球—地质学]