论文部分内容阅读
以阿姆河盆地右岸地区碳酸盐岩台内滩气田为例,运用基于图论的多分辨率聚类分析法开展了以常规测井资料为基础的测井相聚类及岩相定量预测研究。该方法不需要分析数据体的结构及聚类数等先验知识为指导,能够自动优选最佳聚类个数,并允许按照实际需求控制聚类级别,进行聚类合并。依据岩芯资料岩相识别及测井相标定结果,本文最终建立了包含5个聚类的测井相划分模型及岩相定量解释图版,其中,聚类测井相1~5分别对应于泻湖泥、石膏坪、滩间、低能滩及高能滩,取芯段符合率达85%以上,能够较好的运用于非取芯段岩相预测研究。据此,我们进行了连续的聚类测井相划分及岩相预测,并对层序地层格架内岩相分布及物性特征进行了分析。
Taking the carbonate platform of the Tainitan area on the right bank of the Amu River Basin as an example, the logging facies clustering based on conventional log data and the quantitative prediction of lithofacies were carried out by using multi-resolution cluster analysis based on graph theory the study. This method does not need to analyze the structure of the data body and the number of clusters and other prior knowledge as a guide, can automatically optimize the optimal number of clusters, and allows the cluster level control according to the actual needs, cluster consolidation. Based on the lithofacies recognition and logging phase calibration results, a logging facies model and lithofacies interpretation map with five clusters are finally established. The clustering logs 1 ~ 5 correspond to the lagoons Mud, gypsum plaster, beach, low energy beach and high energy beach. The coincidence rate of coring section is over 85%, which can be applied to the prediction of non-coring section lithofacies well. Based on this, we conducted continuous clustering logging facies division and lithofacies prediction, and analyzed the lithofacies distribution and physical characteristics in the sequence stratigraphic framework.