应用最小二乘支持向量机识别广利油田沙四段储层岩性  

A lithologic identification method in Es4reservoir of Guangli oilfield with Least square support vector machine

作　　者：韩学辉[1] 支乐菲[2] 刘荣 杨体源[4] 李亚萍[4] 

机构地区：[1]中国石油大学(华东)地球科学与技术学院,青岛266580 [2]胜利油田东胜精攻石油开发集团股份有限公司,东营257000 [3]中国石油集团海洋工程有限公司天津分公司,塘沽300451 [4]中国石油天然气股份有限公司青海油田分公司勘探开发研究院,敦煌736200

出　　处：《地球物理学进展》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

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.

关 键 词：最小二乘支持向量机 岩性 测井 非均质性 

分 类 号：P631[天文地球—地质矿产勘探]