石油与天然气地质 ›› 2022, Vol. 43 ›› Issue (5): 1259-1270.doi: 10.11743/ogg20220520

• 方法技术 • 上一篇    下一篇

基于多元统计分析的油-源对比

郭喜浩1(), 徐昉昊1(), 黄晓波2, 江涛2, 梁浩然1, 李长志1, 李智超1   

  1. 1.成都理工大学 油气藏地质及开发工程国家重点实验室,四川 成都 610059
    2.中海石油(中国)有限公司 天津分公司,天津 300459
  • 收稿日期:2021-09-24 修回日期:2022-07-10 出版日期:2022-10-01 发布日期:2022-09-02
  • 通讯作者: 徐昉昊 E-mail:419312387@qq.com;xufanghao17@cdut.edu.cn
  • 第一作者简介:郭喜浩(1997—),男,硕士研究生,油气藏地质学与成藏动力学。E?mail: 419312387@qq.com
  • 基金项目:
    国家科技重大专项(2016ZX05024-002)

Oil-source correlation based on multivariate statistical analysis: A case study of the Bodong Sag, Bohai Bay Basin

Xihao Guo1(), Fanghao Xu1(), Xiaobo Huang2, Tao Jiang2, Haoran Liang1, Changzhi Li1, Zhichao Li1   

  1. 1.State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation,Chengdu University of Technology,Chengdu,Sichuan 610059,China
    2.Tianjin Branch of CNOOC,Tianjin 300459,China
  • Received:2021-09-24 Revised:2022-07-10 Online:2022-10-01 Published:2022-09-02
  • Contact: Fanghao Xu E-mail:419312387@qq.com;xufanghao17@cdut.edu.cn

摘要:

多元统计分析可以综合考虑多个相关参数间的相互联系,在分类判别方面具有独特的优势,是多套烃源岩发育的复杂背景下油气源对比的有效方法。然而不同的多元统计方法在同一案例中的对比应用以及各自的适用性尚没有学者进行专门的研究分析。渤海湾盆地渤东凹陷发育多套烃源岩,不同构造的油气来源异常复杂,是利用多元统计分析进行油-源对比的最佳案例。综合对比主成分分析、Fisher线性判别分析以及Bayes线性判别分析的应用情况,发现主成分分析不能区分不同层位的烃源岩,因而不能用于渤东凹陷的油源对比,而Fisher和Bayes线性判别分析建立的模型可以区分不同层位的烃源岩,但都不能准确识别混源油。在Bayes判别模型得到的分类概率的基础上,通过优化Bayes判别模型的判别准则,建立了适用于渤东凹陷的油源判别模型,并识别出渤东凹陷存在4类油族,其中油族A、油族B、油族C分别来源于东营组三段(东三段)、沙一段-沙二段、沙三段烃源岩,而油族D为来源于东三段和沙一段-沙二段烃源岩的混源油。

关键词: 判别模型, 线性判别分析, 多元统计分析, 油-源对比, 渤东凹陷, 渤海湾盆地

Abstract:

Multivariate statistical analysis can comprehensively consider the interrelationship of multiple related parameters, and is of unique advantage in classification-based identification. It is thus an effective method for oil-source correlation with multiple suites of source rocks occurring simultaneously in a zone. However, no authors have conducted special research and analysis on the comparative application and applicability of diverse multivariate statistical methods to one case. There are multiple suites of source rocks developed in the Bodong Sag, Bohai Bay Basin, and the sources of oil and gas in different structures are extremely complex, which is the best case for oil-source correlation using multivariate statistical analysis. The comparison of principal component analysis (PCA), Fisher linear discriminant analysis (LDA) and Bayes LDA in terms of application shows that PCA cannot be applied to distinguish source rocks at different horizons, so it cannot be used for oil-source correlation in the Bodong Sag. However, the models established by Fisher and Bayes LDA can do this effectively, though neither of them could accurately identify oil of mixed sources. Based on the classification probability obtained from the Bayes LDA model, a discriminant model suitable for the Bodong Sag is established with the discriminant criteria of the Bayes model optimized, and four oil familys are identified in the Bodong Sag, with oil family A, B and C are derived from the source rocks of the third member of the Oligocene Dongying Formation (Ed3), the first to second member (Es1-2) and the third member (Es3) of the Eocene Shahejie Formation, respectively, and oil family D from the source rocks of Ed3and Es1-2.

Key words: discriminant model, linear discriminant analysis (LDA), multivariate statistical analysis, oil-source correlation, Bodong Sag, Bohai Bay Basin

中图分类号: