应用最小二乘支持向量机识别J13井区杜家台油层岩性  

Identification of the lithology of Dujiatai formation in J13 well block with least squares support vector machine

作　　者：韩学辉[1] 卢时林[2] 支乐菲[1] 刘贵满[2] 廖永斌[2] 

机构地区：[1]中国石油大学,山东青岛266555 [2]中油辽河油田公司,辽宁凌海121209

出　　处：《特种油气藏》2011年第6期18-21,124,共4页Special Oil & Gas Reservoirs

基　　金：山东省自然科学基金项目"含油气砂岩储层岩石物理实验快速测量方法"(Y2008E08)

摘　　要：J13井区杜家台油层的储层岩性以低孔低渗中砂岩、细砂岩和不等粒砂岩为主,常规交会图法难以识别,无法确保按岩性建立的测井评价模型的配套使用。为此,应用最小二乘支持向量机对储层岩性进行了识别。首先分析了最小二乘支持向量机的方法原理和实现流程;其次随机选取了J13井区部分层段的岩性和测井响应建立了学习样本集,采用网格搜索法确定了参数C和δ,自检结果表明识别正确率可达到94%;最后对J13井区储层岩性做了测井识别,与岩心分析资料进行对比,岩性识别正确率达到87.5%。研究表明,最小二乘支持向量机可满足J13井区岩性识别的需要。Dujiatai formation in J13 well block is a low porosity and low permeability reservoir made up of medium granular sandstone,fine grained sandstone and inequigranular sandstone,whose lithology is difficult to be identified with conventional crossplot method,so there is no guarantee for the use of the log evaluation model set up based on lithology.This paper used the least squares support vector machine to identify the lithology of the formation.First,the principle and the method were introduced;then the lithology and logging responses from some sections of J13 well block were randomly selected to set up a collection of learning samples;the parameter C and were determined using the grid search method;an accuracy rate of identification up to 94% was indicated by self-check results.Finally,the lithology of the formation in J13 well block was identified by logging and compared with core analysis data,reaching an accuracy rate of lithology identification of 87.5%.The study indicates that the least squares support vector machine can satisfy the needs of identifying the lithology in J13 well block.

关 键 词：最小二乘支持向量机 低孔低渗 岩性 测井响应 J13井区 

分 类 号：TE122.1[石油与天然气工程—油气勘探] TE142