映射分析中的自组织方法在烃类检测中的应用

Self-organizing mapping analysis of petroleum detection

自组织分析是在自组织映射神经网络基础上发展的一种新的非监督聚类和数据分析方法 ,其核心是将神经网络学习所用原数据体映射为自组织映射密度图。在自组织映射密度图上 ,原空间中具有较大的相似性的数据样本被映射到同一个位置或相近的两个位置。勘探早期可用于约束储集层预测的井很少 ,借助于原数据体中存在的聚类关系 ,利用自组织分析技术对数据体隐含的油气信息进行聚类分析 ,有助于提高钻井成功率。

Self organizing mapping analysis of reservoir study is a new technique developed through self organizing mapping nerve network, whose major idea is the self organizing mapping density map obtained from nerve network learning. Density map is a map of the data body used in nerve network learning. It maps two same or close samples in the data body to a position or two close ones, which shows that these two samples have strong comparability in original space. There are few well locations used to restrain reservoir prediction at the early stage of exploration. Therefore, the authors of this paper had to resort to the clustering relationship existing in original data body, apply self organizing mapping technology to the performance of the clustering analysis on the implicit oil and gas information of the data body and locate the well finally so as to improve the success ratio of future drilling greatly.