基于随机森林的致密储层分类——以延安气田东部盒8段为例  

Classification of Tight Reservoirs Based on Random Forest Algorithm:A Case Study of He 8 Member in East of Yan’an Gasfield

作　　者：王妍[1] 王若谷 魏克颖[3] 卢海娇 熊小伟 张天杰 WANG Yan;WANG Ruogu;WEI Keying;LU Haijiao;XIONG Xiaowei;ZHANG Tianjie(College of Petroleum Engineering,Xi’ an Shiyou University,Xi’ an,Shaanxi 710065,China;Research Institute,Shaanxi Yanchang Petroleum (Group) Co.,Ltd.,Xi’ an,Shaanxi 710006,China;No.3 Gas Production Plant,PetroChina Changqing Oilfield Company,Wushenqi,Inner Mongolia 017300,China;Changbei Operating Company,PetroChina Changqing Oilfield Company,Xi’ an,Shaanxi 710000,China;No.3 Oil Production Plant,PetroChina Changqing Oilfield Company,Yinchuan,Ningxia 750000,China;No.1 Oil Production Plant,PetroChina Changqing Oilfield,Xi’ an,Shaanxi 710000,China)

机构地区：[1]西安石油大学石油工程学院,陕西西安710065 [2]陕西延长石油(集团)有限责任公司研究院,陕西西安710006 [3]中国石油长庆油田分公司第三采气厂,内蒙古乌审旗017300 [4]中国石油长庆油田分公司长北作业分公司,陕西西安710000 [5]中国石油长庆油田分公司第三采油厂,宁夏银川750000 [6]中国石油长庆油田分公司第一采油厂,陕西西安710000

出　　处：《西安石油大学学报（自然科学版）》2021年第6期1-8,共8页Journal of Xi’an Shiyou University（Natural Science Edition）

基　　金：陕西省自然科学基础研究计划项目(2017JQ4007)。

摘　　要：为了解决储层分类涉及到指标多、需要根据个人经验判别、耗时耗力的问题,提出一种基于随机森林的致密储层分类方法。首先,选取延安气田东部下石盒子组盒8段致密砂岩储层高压压汞曲线进行研究,利用主成分分析法降维得到用于分类的4种特征。再采用基于CART决策树的随机森林算法进行分类,得到储层分类结果。最后,对分类结果进行校验并优化随机森林算法的参数。实验结果表明,所提出的储层分类方法准确率高,计算代价小,能有效降低时间损耗、节省人力成本。A tight reservoir classification method based on random forest algorithm is proposed in order to solve the problem that reservoir classification involves many indicators and requires a lot of time and manpower to distinguish reservoir types according to personal experience.Firstly,the high-pressure mercury injection curves of tight sandstone reservoirs in He 8 member of Lower Shihezi Formation in the east of Yan’an gas field are studied,and four characteristics for classification are obtained by principal component analysis.Then they are used for classification,and the reservoir classification results are obtained using the random forest algorithm based on CART decision tree.Finally,the classification results are verified and the parameters of random forest algorithm are optimized.The experimental results show that the proposed reservoir classification method has high accuracy and small amount of calculation,and can effectively reduce time and labor cost for calculation.

关 键 词：储层分类 随机森林 主成分分析 致密砂岩 

分 类 号：TE19[石油与天然气工程—油气勘探]