主成分分析法在致密砂岩岩性识别的应用研究
Application research of principal component analysis method in lithology identification of tight sandstone
浏览(1012) 下载(8)
- 引用格式:
-
王宵宇,谢然红,毛治国,吴勃翰,徐陈昱,卫弘媛.主成分分析法在致密砂岩岩性识别的应用研究[J].天然气与石油,2021,39(1):88-93.doi:10.3969/j.issn.1006-5539.2021.01.014
WANG Xiaoyu, XIE Ranhong, MAO Zhiguo, WU Bohan, XU Chenyu, WEI Hongyuan.Application research of principal component analysis method in lithology identification of tight sandstone[J].Natural Gas and Oil,2021,39(1):88-93.doi:10.3969/j.issn.1006-5539.2021.01.014
- DOI:
- 10.3969/j.issn.1006-5539.2021.01.014
- 作者:
- 王宵宇1,2，谢然红1,2，毛治国3，吴勃翰1,2，徐陈昱1,2，卫弘媛1,2
WANG Xiaoyu1,2, XIE Ranhong1,2, MAO Zhiguo3, WU Bohan1,2, XU Chenyu1,2, WEI Hongyuan1,2
- 作者单位:
- 1. 中国石油大学(北京)油气资源与探测国家重点实验室,2. 中国石油大学(北京)地球探测与信息技术北京市重点实验室, 3. 中国石油勘探开发研究院
1. State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum (Beijing), Beijing, 102249, China; 2. Beijing Key Laboratory of Earth Exploration and Information Technology, China University of Petroleum(Beijing), Beijing, 102249, China; 3. Research Institute of Petroleum Exploration and Development, PetroChina, Beijing, 100083, China
- 关键词:
- 致密砂岩储层;测井解释;岩性识别;主成分分析法;交会图法
Tight sandstone reservoirs; Logging interpretation; Lithology identification; Principal component analysis method; Crossplot method
- 摘要:
- 致密砂岩储层孔隙结构复杂，非均质性强,岩性识别困难。传统的基于特征曲线进行曲线重叠、构造参数或建立交会图识别岩性的方法依赖解释人员的知识和经验,而机器学习方法基于测井曲线和岩性类别的映射关系进行数理统计,无法直接观察岩性识别的过程。因此,提出利用主成分分析法来对致密砂岩岩性进行识别,选取对研究区岩性敏感的自然电位、补偿中子、密度、声波时差和核磁共振横向弛豫时间分布T 2几何均值曲线作为主成分分析法的输入变量,提取累计贡献率为91%的主成分F 1和F 2建立交会图识别岩性。选取鄂尔多斯盆地姬塬地区的岩心分析资料和测井资料对主成分分析法的识别结果进行验证,结果表明：F 1—F 2交会图有效划分含砾粗砂岩、中砂岩、细砂岩和泥岩；对于砂岩的含油级别,交会图能有效划分荧光细砂岩、油迹细砂岩和油斑细砂岩。通过应用实例分析表明,主成分分析法对致密砂岩岩性的识别具有可行性。
Due to the complex pore structure and strong heterogeneity in tight sandstone reservoirs, it is difficult to identify the lithology. Traditional method makes use of characteristic curves to overlap curves, develop parameters, or establish crossplots to identify lithology, which relies on the knowledge and experience of interpreters. While the machine learning methods perform mathematical statistics based on the mapping relationship between the logs and the lithology, but the lithology recognition process can not directly be observed. Therefore, the principal component analysis method is used to identify the lithology of the tight sandstone. The logs of spontaneous potential, compensated neutron, density, acoustic time difference, and T 2 geometric mean of NMR transverse relaxation time distribution that are sensitive to lithology in the study area are selected as the input variables of the principal component analysis method. The principal components F 1 and F 2 with a cumulative contribution rate of 91% are extracted to establish a crossplot to identify lithology. The core analysis data and well log data of Jiyuan area in Ordos Basin are selected to verify the identification results of the principal component analysis method. The results show that the F 1-F 2 crossplot can effectively divide the gravel coarse sandstone, medium sandstone, fine sandstone, and mudstone. Also, for the oil grade of sandstone, the crossplot can effectively divides fluorescent fine sandstone, oil trace fine sandstone, and oil spot fine sandstone. The application case shows that the principal component analysis method is feasible to identify the lithology of tight sandstone.
