Prediction of petroleum resource abundance based on artificial neural network method: a case study of third member of Paleogene Shahejie Formation in Wenliu area of Dongpu Sag, Bohai Bay Basin
-
摘要: 油气资源丰度通常受多个因素控制，其相关参数信息种类繁杂、数据量庞大，应用传统的地质统计学方法定量预测准确度不高。为了快速预测油气资源量丰度并明确其主控因素，以渤海湾盆地东濮凹陷文留地区古近系沙河街组三段为例，采用基于多层感知器神经网络(MLP)方法对油气资源丰度进行定量预测，同时采用Boosting集成学习算法优化预测模型，分别对66组样本油气资源丰度数据进行预测。结果表明，训练集数据实测值与预测值相关系数分别达0.789和0.989，验证集数据实测值与预测值相关系数分别达0.618和0.825，测试数据中实测值和预测值相关系数分别达0.689和0.845；有效厚度、平均渗透率、有效孔隙度是影响油气资源丰度最主要的3个地质因素，重要性系数分别为33.93%、20.12%和19.53%，圈闭面积、地面原油密度、生烃中心贡献等参数为次要影响因素。采用Boosting集成学习算法优化之后的多层感知器模型预测准确性得到了很大的提升，能为有利目标优选及勘探开发方案调整提供可靠依据，为凹陷内其他区块油气资源评价提供较好的参考和借鉴。Abstract: The abundance of petroleum resource is influenced by various factors and involves complex parameters and extensive data. Consequently, traditional geostatistical methods often lack precision in quantitative prediction. To address this issue, this study focuses on the third member of Paleogene Shahejie Formation (member Es3) in the Wenliu area of the Dongpu Sag and utilizes a multi-layer perceptron neural network (MLP) for predicting petroleum resource abundance and employed the Boosting ensemble learning algorithm to optimize the prediction model. The MLP and MLP-Boosting algorithm models were test on 66 sample groups, yielding correlation coefficients of 0.789 and 0.989 for the training set, 0.618 and 0.825 for the validation set and 0.689 and 0.845 for the test set. The analysis identified effective thickness, average permeability and effective porosity are the most significant geological factors influencing petroleum resource abundance, with importance coefficients of 33.93%, 20.12% and 19.53%, respectively. Other factors such as trap area, surface crude oil density and sedimentary facies assignment were found to be less influential. Overall, the Boosting ensemble learning algorithm significantly enhanced the prediction accuracy of the multi-layer perceptron model, offering valuable insights for target optimization, exploration planning and petroleum resource evaluation in other blocks in the sag.
-
Key words:
- machine learning /
- neural network /
- prediction model /
- resource abundance /
- Dongpu Sag /
- Bohai Bay Basin
-
图 1 渤海湾盆地东濮凹陷研究区位置(a)、沙三段构造等值线(b)和典型油藏剖面(c)
据中国石化中原油田分公司。
Figure 1. Location of study area (a), isolines of third member of Shahejie Formation (b) and typical reservoir sections (c) in Dongpu Sag, Bohai Bay Basin
图 2 渤海湾盆地东濮凹陷新生界地层柱状图
Figure 2. Stratigraphic histogram of Cenozoic in Dongpu Sag, Bohai Bay Basin
图 3 渤海湾盆地东濮凹陷研究区排烃强度及烃源岩贡献赋值综合示意
据参考文献[25]修改。
Figure 3. Comprehensive schematic diagram of hydrocarbon expulsion intensity and contribution assignments of hydrocarbon source rocks in study area in Dongpu Sag, Bohai Bay Basin
图 4 渤海湾盆地东濮凹陷文留地区沙三段沉积相赋值综合示意
据中国石化中原油田分公司。
Figure 4. Comprehensive schematic diagram of sedimentary facies assignment in third member of Shahejie Formation in Wenliu area of Dongpu Sag, Bohai Bay Basin
图 5 多层感知器学习网络结构示意
Figure 5. Structure of multi-layer perceptron (MLP) learning network
图 8 基于多层感知器神经网络油气资源丰度建模结果
Figure 8. Results of petroleum resource abundance modeling based on MLP neural network
图 9 基于MLP-Boosting算法油气资源丰度建模结果
Figure 9. Results of petroleum resource abundance modeling based on MLP-Boosting algorithm modeling
图 10 MLP模型(a)和MLP-Boosting集成算法模型(b)检验数据相关性
Figure 10. Correlation coefficients of checking data of MLP model (a) and MLP-Boosting ensemble algorithm model (b)
图 11 MLP模型和MLP-Boosting集成算法模型可靠性分析
Figure 11. Reliability of MLP and MLP-Boosting ensemble algorithm models
图 12 渤海湾盆地东濮凹陷古近系沙河街组三段预选有利区分布
Figure 12. Distribution of preselected favorable areas of third member of Paleogene Shahejie Formation in Dongpu Sag, Bohai Bay Basin
图 13 渤海湾盆地东濮凹陷古近系沙河街组三段地质参数相关系数矩阵热图
Figure 13. Matrix heat map of correlation coefficients of geological parameters in third member of Paleogene Shahejie Formation, Dongpu Sag, Bohai Bay Basin
图 14 渤海湾盆地东濮凹陷古近系沙河街组三段地质参数数据
Figure 14. Geological parameters of third member of Paleogene Shahejie Formation, Dongpu Sag, Bohai Bay Basin
表 1 渤海湾盆地东濮凹陷文留地区沉积相赋值统计
Table 1. Statistics of sedimentary facies assignments of Wenliu area of Dongpu Sag, Bohai Bay Basin
参数 滨浅湖 滨浅湖—三角洲前缘 三角洲前缘 半深湖—深湖 分流河道 浅水三角洲 地质储量/104 t 319 118 1302 484.87 124 449 地质储量占比 0.11 0.04 0.47 0.17 0.04 0.16 沉积相赋值 0.25 0.09 1.00 0.37 0.10 0.34 
下载: 导出CSV
表 2 基于多层感知器神经网络模型油气资源丰度预测结果
Table 2. Prediction results of petroleum resource abundance based on MLP neural network modeling
下载: 导出CSV
表 3 基于MLP-Boosting算法模型油气资源丰度预测结果
Table 3. Prediction results of petroleum resource abundance based on MLP-Boosting algorithm model
参数 训练集 验证集 最小误差/(104 t/km2) -6.627 -21.478 最大误差/(104 t/km2) 5.963 44.258 平均误差/(104 t/km2) -0.205 1.754 平均绝对误差/(104 t/km2) 1.287 11.399 标准差 2.068 16.545 相关系数 0.989 0.825 样品数 47 19
下载: 导出CSV
表 4 MLP模型和MLP-Boosting集成算法模型检验结果
Table 4. Test results of MLP and MLP-Boosting ensemble algorithm models
参数 检验数据 MLP MLP-Boosting 最小误差/(104 t/km2) -14.362 -11.288 最大误差/(104 t/km2) 38.262 36.32 平均误差/(104 t/km2) 7.6 3.507 平均绝对误差/(10 4t/km2) 11.408 8.936 相关系数 0.689 0.845 样品数 20 20
下载: 导出CSV
表 5 渤海湾盆地东濮凹陷古近系沙河街组三段3个预选有利区带的资源丰度预测结果
Table 5. Resource abundance evaluation results of three pre-selected favorable zones in third member of Paleogene Shahejie Formation of Dongpu Sag, Bohai Bay Basin
预选有利区带 有利区带资源丰度/(104 t/km2) MLP- Boosting算法模型 MLP模型 最大值 最小值 平均值 最大值 最小值 平均值 区带1 75.79 47.11 61.37 64.84 33.20 59.13 区带2 65.27 36.28 56.16 65.43 35.03 55.23 区带3 65.32 65.12 65.21 65.43 38.56 63.54
下载: 导出CSV
-
[1] 柳广弟, 刘成林, 郭秋麟. 油气资源评价[M]. 北京: 石油工业出版社, 2018.LIU Guangdi, LIU Chenglin, GUO Qiulin. Oil and gas resources evaluation[M]. Beijing: Petroleum Industry Press, 2018. [2] 张蔚, 刘成林, 吴晓智, 等. 中国不同类型盆地油气资源丰度统计特征及预测模型[J]. 地质与勘探, 2019, 55(6): 1518-1527. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKT201906018.htmZHANG Wei, LIU Chenglin, WU Xiaozhi, et al. Statistical characteristics and prediction models for oil and gas resources abundance in different types of Chinese basins[J]. Geology and Exploration, 2019, 55(6): 1518-1527. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKT201906018.htm [3] 鄢琦, 周总瑛. 中国东部断陷盆地石油资源丰度统计模型的建立[J]. 石油实验地质, 2009, 31(3): 292-295. doi: 10.11781/sysydz200903292YAN Qi, ZHOU Zongying. Establishment on oil resources abundance statistical model in East China rift basins[J]. Petroleum Geology & Experiment, 2009, 31(3): 292-295. doi: 10.11781/sysydz200903292 [4] 赵文智, 胡素云, 王红军, 等. 中国中低丰度油气资源大型化成藏与分布[J]. 石油勘探与开发, 2013, 40(1): 1-13. https://www.cnki.com.cn/Article/CJFDTOTAL-SKYK201301002.htmZHAO Wenzhi, HU Suyun, WANG Hongjun, et al. Large-scale accumulation and distribution of medium-low abundance hydrocarbon resources in China[J]. Petroleum Exploration and Deve-lopment, 2013, 40(1): 1-13. https://www.cnki.com.cn/Article/CJFDTOTAL-SKYK201301002.htm [5] 张金川, 金之钧, 郑浚茂. 深盆气资源量-储量评价方法[J]. 天然气工业, 2001, 21(4): 32-35. https://www.cnki.com.cn/Article/CJFDTOTAL-TRQG200104008.htmZHANG Jinchuan, JIN Zhijun, ZHENG Junmao. Deep basin gas resource-reserve evaluation method[J]. Natural Gas Industry, 2001, 21(4): 32-35. https://www.cnki.com.cn/Article/CJFDTOTAL-TRQG200104008.htm [6] 崔宝文, 赵莹, 张革, 等. 松辽盆地古龙页岩油地质储量估算方法及其应用[J]. 大庆石油地质与开发, 2022, 41(3): 14-23. https://www.cnki.com.cn/Article/CJFDTOTAL-DQSK202203002.htmCUI Baowen, ZHAO Ying, ZHANG Ge, et al. Estimation method and application for OOIP of Gulong shale oil in Songliao Basin[J]. Petroleum Geology & Oilfield Development in Daqing, 2022, 41(3): 14-23. https://www.cnki.com.cn/Article/CJFDTOTAL-DQSK202203002.htm [7] 郭秋麟, 武娜, 闫伟, 等. 深层天然气资源评价方法[J]. 石油学报, 2019, 40(4): 383-394. https://www.cnki.com.cn/Article/CJFDTOTAL-SYXB201904001.htmGUO Qiulin, WU Na, YAN Wei, et al. An assessment method for deep gas resources[J]. Acta Petrolei Sinica, 2019, 40(4): 383-394. https://www.cnki.com.cn/Article/CJFDTOTAL-SYXB201904001.htm [8] 徐旭辉, 周卓明, 宋振响, 等. 油气资源评价方法关键参数研究和资源分布特征: 以中国石化探区"十三五"资源评价为例[J]. 石油实验地质, 2023, 45(5): 832-843. doi: 10.11781/sysydz202305832XU Xuhui, ZHOU Zhuoming, SONG Zhenxiang, et al. Methods and key parameters for oil and gas resource assessment and distribution characteristics of oil and gas resource: a case study of resource assessment of SINOPEC during the 13th Five-Year Plan period[J]. Petroleum Geology & Experiment, 2023, 45(5): 832-843. doi: 10.11781/sysydz202305832 [9] 凡玉梅. 未开发油气储量不确定性潜力评价方法[J]. 石油实验地质, 2022, 44(6): 1100-1104. doi: 10.11781/sysydz2022061100FAN Yumei. Evaluation method for uncertain potential of undeveloped reserves[J]. Petroleum Geology & Experiment, 2022, 44(6): 1100-1104. doi: 10.11781/sysydz2022061100 [10] 柳广弟, 胡素云, 赵文智. 中国主要含油气盆地运聚单元石油资源丰度及其预测模型[J]. 石油勘探与开发, 2006, 33(6): 759-761. https://www.cnki.com.cn/Article/CJFDTOTAL-SKYK200606021.htmLIU Guangdi, HU Suyun, ZHAO Wenzhi. Oil resource abundance of petroleum plays in Chinese basins and its prediction model[J]. Petroleum Exploration and Development, 2006, 33(6): 759-761. https://www.cnki.com.cn/Article/CJFDTOTAL-SKYK200606021.htm [11] 白琨琳, 赵迎冬. 油气资源评价中成因法分析与运聚系数取值模型研究[J]. 地质与勘探, 2021, 57(3): 656-666. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKT202103019.htmBAI Kunlin, ZHAO Yingdong. Valuation model of the migration-accumulation coefficient in the genetic method for assessment of oil and gas resources[J]. Geology and Exploration, 2021, 57(3): 656-666. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKT202103019.htm [12] GUO Qiulin, REN Hongjia, WU Xiaozhi, et al. A fractal simulation method for simulating the resource abundance of oil and gas and its application[J]. Mathematical Geosciences, 2022, 54(5): 873-901. [13] 姚纪明, 于炳松, 车长波, 等. 组合法在塔里木盆地石油产量预测中的应用[J]. 自然资源学报, 2009, 24(5): 907-914. https://www.cnki.com.cn/Article/CJFDTOTAL-ZRZX200905019.htmYAO Jiming, YU Bingsong, CHE Changbo, et al. Application of a combination forecast model in the trend forecast of oil production in the Tarim Basin[J]. Journal of Natural Resources, 2009, 24(5): 907-914. https://www.cnki.com.cn/Article/CJFDTOTAL-ZRZX200905019.htm [14] ZHANG Guoyin, WANG Zhizhang, LI Huaji, et al. Permeability prediction of isolated channel sands using machine learning[J]. Journal of Applied Geophysics, 2018, 159: 605-615. [15] 刘铁桩, 张清正, 李亚萍. 东濮凹陷天然气资源潜力、勘探方向及前景[J]. 西部探矿工程, 2001, 13(5): 45-46. https://www.cnki.com.cn/Article/CJFDTOTAL-XBTK200105024.htmLIU Tiezhu, ZHANG Qingzheng, LI Yaping. The natural gas resource potential and exploration direction and looking forward of Dongpu Sag[J]. West-China Exploration Engineering, 2001, 13(5): 45-46. https://www.cnki.com.cn/Article/CJFDTOTAL-XBTK200105024.htm [16] 董潇阳. 东濮凹陷油气资源潜力与分布评价[D]. 北京: 中国石油大学(北京), 2018.DONG Xiaoyang. Potential and distribution of oil and gas resources in Dongpu Sag[D]. Beijing: China University of Petroleum (Beijing), 2018. [17] 谈玉明, 李红磊, 张云献, 等. 东濮凹陷古近系优质烃源岩特征与剩余资源潜力分析[J]. 断块油气田, 2020, 27(5): 551-555. https://www.cnki.com.cn/Article/CJFDTOTAL-DKYT202005003.htmTAN Yuming, LI Honglei, ZHANG Yunxian, et al. Analysis to high quality source rock characteristics and residual resource potential in Dongpu Sag in Paleogene[J]. Fault-Block Oil & Gas Field, 2020, 27(5): 551-555. https://www.cnki.com.cn/Article/CJFDTOTAL-DKYT202005003.htm [18] 余海波, 程秀申, 徐田武, 等. 东濮凹陷古近系构造特征及其对油气成藏的控制作用[J]. 油气地质与采收率, 2021, 28(3): 42-52. https://www.cnki.com.cn/Article/CJFDTOTAL-YQCS202103006.htmYU Haibo, CHENG Xiushen, XU Tianwu, et al. Paleogene tectonic characteristics and their controlling effect on hydrocarbon accumulation in Dongpu Sag[J]. Petroleum Geology and Reco-very Efficiency, 2021, 28(3): 42-52. https://www.cnki.com.cn/Article/CJFDTOTAL-YQCS202103006.htm [19] 胡斌, 陈传浩, 王长征, 等. 东濮凹陷文留地区沙三中(Es3中)遗迹化石与沉积环境[J]. 河南理工大学学报(自然科学版), 2017, 36(3): 40-46. https://www.cnki.com.cn/Article/CJFDTOTAL-JGXB201703007.htmHU Bin, CHEN Chuanhao, WANG Changzheng, et al. Trace fossils and sedimentary environments in the middle part of third member Shahejie Formation in Wenliu area, Dongpu Sag[J]. Journal of Henan Polytechnic University (Natural Science), 2017, 36(3): 40-46. https://www.cnki.com.cn/Article/CJFDTOTAL-JGXB201703007.htm [20] 慕小水. 东濮凹陷文留地区含盐层系油气成藏机理与模式[D]. 北京: 中国地质大学(北京), 2011.MU Xiaoshui. Hydrocarbon reservoir formation mechanism and pattern for saline series in Wenliu area, Dongpu Depression[D]. Beijing: China University of Geosciences (Beijing), 2011. [21] 刘宣威, 王学军, 李红磊, 等. 东濮凹陷古近系烃源岩特征及其形成环境分析[J]. 断块油气田, 2021, 28(4): 452-455. https://www.cnki.com.cn/Article/CJFDTOTAL-DKYT202104005.htmLIU Xuanwei, WANG Xuejun, LI Honglei, et al. Characteristics and formation environment analysis of Paleogene source rocks in Dongpu Depression[J]. Fault-Block Oil & Gas Field, 2021, 28(4): 452-455. https://www.cnki.com.cn/Article/CJFDTOTAL-DKYT202104005.htm [22] 李浩, 王保华, 陆建林, 等. 东濮凹陷古近系页岩油富集地质条件与勘探前景[J]. 中国石油大学学报(自然科学版), 2021, 45(3): 33-41. https://www.cnki.com.cn/Article/CJFDTOTAL-SYDX202103004.htmLI Hao, WANG Baohua, LU Jianlin, et al. Geological characteristics and exploration prospects of Paleogene continental shale oil accumulation in Dongpu Sag, Bohai Bay Basin[J]. Journal of China University of Petroleum (Edition of Natural Science), 2021, 45(3): 33-41. https://www.cnki.com.cn/Article/CJFDTOTAL-SYDX202103004.htm [23] 高晓红, 李兴奇. 主成分分析中线性无量纲化方法的比较研究[J]. 统计与决策, 2020, 36(3): 33-36. https://www.cnki.com.cn/Article/CJFDTOTAL-TJJC202003007.htmGAO Xiaohong, LI Xingqi. Comparative study on linear dimensionless methods in principal component analysis[J]. Statistics & Decision, 2020, 36(3): 33-36. https://www.cnki.com.cn/Article/CJFDTOTAL-TJJC202003007.htm [24] 郭秋麟, 任洪佳, 于京都, 等. 基于贝叶斯网络的油气勘探风险预测方法: 以准噶尔盆地腹部侏罗系三工河组为例[J]. 中国石油勘探, 2023, 28(1): 108-119. https://www.cnki.com.cn/Article/CJFDTOTAL-KTSY202301010.htmGUO Qiulin, REN Hongjia, YU Jingdu, et al. Prediction method of petroleum exploration risks based on Bayesian network: a case study of the Jurassic Sangonghe Formation in the hinterland of Junggar Basin[J]. China Petroleum Exploration, 2023, 28(1): 108-119. https://www.cnki.com.cn/Article/CJFDTOTAL-KTSY202301010.htm [25] 姜福杰, 姜振学, 庞雄奇. 东营凹陷油气成藏体系的划分及定量评价[J]. 地球科学(中国地质大学学报), 2008, 33(5): 651-660. https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX200805011.htmJIANG Fujie, JIANG Zhenxue, PANG Xiongqi. Division and quantitative evaluation of petroleum accumulation system in Dongying Sag[J]. Earth Science(Journal of China University of Geosciences), 2008, 33(5): 651-660. https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX200805011.htm [26] ASSI K J, NAHIDUZZAMAN K M, RATROUT N T, et al. Mode choice behavior of high school goers: evaluating logistic regression and MLP neural networks[J]. Case Studies on Transport Policy, 2018, 6(2): 225-230. [27] KEARNS M J, VALIANT L G. Learning Boolean formulae or finite automata is as hard as factoring[R]. Cambridge: Harvard University, Center for Research in Computing Technology, Aiken Computation Laboratory, 1988. [28] 陈凯, 朱钰. 机器学习及其相关算法综述[J]. 统计与信息论坛, 2007, 22(5): 105-112. https://www.cnki.com.cn/Article/CJFDTOTAL-TJLT200705022.htmCHEN Kai, ZHU Yu. A summary of machine learning and related algorithms[J]. Statistics & Information Forum, 2007, 22(5): 105-112. https://www.cnki.com.cn/Article/CJFDTOTAL-TJLT200705022.htm [29] 吕传炳, 庞雄奇, 马奎友, 等. 渤海湾盆地束鹿凹陷"牙刷状"油藏成藏特征与模式[J]. 石油与天然气地质, 2022, 43(3): 566-581. https://www.cnki.com.cn/Article/CJFDTOTAL-SYYT202203007.htmLÜ Chuanbing, PANG Xiongqi, MA Kuiyou, et al. Characteristics and reservoiring patterns of "teeth-brush-shaped" oil pools in the Shulu Sag, Bohai Bay Basin[J]. Oil & Gas Geology, 2022, 43(3): 566-581. https://www.cnki.com.cn/Article/CJFDTOTAL-SYYT202203007.htm [30] 张晓磊, 唐颖, 李卓奕, 等. 鄂尔多斯盆地西南缘延长组长8段低充注油藏成藏模式: 以环西-彭阳地区为例[J]. 西安石油大学学报(自然科学版), 2023, 38(1): 31-44. https://www.cnki.com.cn/Article/CJFDTOTAL-XASY202301004.htmZHANG Xiaolei, TANG Ying, LI Zhuoyi, et al. Hydrocarbon accumulation patterns of Chang 8 low-charging oil reservoirs in southwestern margin of Ordos Basin: taking Huanxi-Pengyang area as an example[J]. Journal of Xi'an Shiyou University (Natural Science Edition), 2023, 38(1): 31-44. https://www.cnki.com.cn/Article/CJFDTOTAL-XASY202301004.htm -
点击查看大图
计量
- 文章访问数: 81
- HTML全文浏览量: 22
- PDF下载量: 16
- 被引次数: 0
苏公网安备32021102000780号