摘要
广利油田沙四段储层岩性以为粉砂岩、细砂岩和不等粒砂岩为主,岩性非均质性较强,应用常规测井交会图识别岩性难度大.本文介绍了最小二乘支持向量机的原理及实现流程,利用网格搜索法确定的参数对δ2(C,δ)=(2000,0.707)开展了广利油田沙四段储层岩性的测井识别.应用效果表明,测井识别岩性与岩心分析资料的符合率达到了86%,可满足广利油田沙四段储层岩性识别的需要.
The lithology in Es4 reservoir of Guangli oilfield is mainly composed of silt sandstone, fine sandstone and unequigranular sandstone. It is difficult to identify lithology with conventional cross-plots. This paper introduces the principle and technique flow of least quares support vector machine(LS-SVM) and presents the detail to distinguish the lithology of Guangli oilfield. Especially, the parameters of C and δ (C, δ) = (2000,0. 707) are gained by grid search method. The application proves that the coincidence rate of the lithologic identification by LS-SVM and core analysis is 86%. So, LS-SVM is able to meet the need of lithology recognization in Es4 reservoir of Guangli oilfield.
出处
《地球物理学进展》
CSCD
北大核心
2013年第4期1886-1892,共7页
Progress in Geophysics
基金
山东省自然科学基金(Y2008E08)
国家油气专项(2011ZX05009-003)联合资助
关键词
最小二乘支持向量机
岩性
测井
非均质性
least squares support vector machine, lithology, well logging, heterogeneity