基于物理经验模型约束的机器学习方法在页岩油产量预测中的应用

周济民; 张海晨; 王沫然

doi:10.21656/1000-0887.420015

基于物理经验模型约束的机器学习方法在页岩油产量预测中的应用

doi: 10.21656/1000-0887.420015

清华大学航天航空学院，北京 100084

基金项目:

国家重点研发项目（2019YFA0708704）

详细信息

作者简介:
周济民(2000—)，男(E-mail: zhoujm18@mails.tsinghua.edu.cn);王沫然(1977—)，男，教授，博士，博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).

通讯作者:
王沫然(1977—)，男，教授，博士，博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).

中图分类号: O368|O29
计量
- 文章访问数: 1206
- HTML全文浏览量: 210
- PDF下载量: 155
- 被引次数: 0
出版历程
- 收稿日期: 2021-01-14
- 修回日期: 2021-04-06
- 网络出版日期: 2021-09-29

Machine Learning With Physical Empirical Model Constraints for Prediction of Shale Oil Production

School of Aerospace Engineering, Tsinghua University, Beijing 100084, P.R.China

摘要

摘要: 页岩油气产量预测是确定其开发经济性的重要手段，目前的产量预测研究很少能在物理模型与数据挖掘方法之间达到统一.针对页岩油气的产量分析，本研究深入结合误差反向传递（BP）神经网络和长短期记忆（LSTM）神经网络的数学方法优势，综合考虑工程经验模型的约束，改善了模型预测精度，经过实例数据训练后可较好地预测油田产量，并研究了页岩储层深度、总有机碳含量（TOC）、脆性度等油田参数对产量预测的影响规律.这项工作可以为页岩油气规模化开发提供可靠的产量预测和经济评价.
- 产量预测 /
- 机器学习 /
- 神经网络 /
- 工程经验模型
Abstract: Prediction of oil and gas production is an important way to determine its development economy. However, at present the production prediction is still hard to achieve consistency between the physics-based method and the data-based method. For shale oil and gas production analysis, in-depth combination of mathematical advantages brought by BP neural networks and LSTM neural networks, and comprehensive consideration of physics-based models, lead to good improvement in the prediction accuracy of the model. After training with practical testing data, the prediction of oilfield production can be significantly improved. Afterwards, the effects of the reservoir depth, the TOC and the brittleness, etc. on production prediction were studied. In conclusion, the work provides reliable production prediction and economic evaluation for large-scale development of shale oil and gas.
- production prediction /
- machine learning /
- neural network /
- engineering empirical model

HTML全文

参考文献(42)

全国煤化工信息总站. 2002年—2018年中国能源生产、消费结构[J]. 煤化工, 2020,48(3): 85.(National Coal Chemical Industry Information Station. China energy production and consumption structure from 2002

to 2018[J].Coal Chemical Industry,2020,48(3): 85.(in Chinese))

[2]徐倩. 新形势下中国石油储备现状及未来规划[J]. 化工管理, 2018(25): 16-17.(XU Qian. The present situation and future planning of China petroleum reserves under the new situation[J].Chemical Enterprise Management,2018(25): 16-17.(in Chinese))

[3]方圆, 张万益, 马芬. 全球页岩油资源分布与开发现状[J]. 矿产保护与利用, 2019,39(5): 126-134.(FANG Yuan, ZHANG Wanyi, MA Fen. Research on the global distribution and development status of shale oil[J].Conservation and Utilization of Mineral Resources,2019,39(5): 126-134.(in Chinese))

[4]高诚, 孙川翔, 苏建政. 国际油页岩开发技术现状及新疆开采可行性分析[J]. 地球科学前沿, 2017,7(3): 330-335.(GAO Cheng, SUN Chuanxiang, SU Jianzheng. Global oil shale development technology and its application analysis in Xinjiang, China[J].Advances in Geosciences,2017,7(3): 330-335.(in Chinese))

[5]杜金虎, 刘合, 马德胜, 等. 试论中国陆相致密油有效开发技术[J]. 石油勘探与开发, 2014,41(2): 198-205.(DU Jinhu, LIU He, MA Desheng, et al. Discussion on effective development techniques for continental tight oil in China[J].Petroleum Exploration & Development,2014,41(2): 198-205.(in Chinese))

[6]张映红, 路保平, 陈作. 中国陆相致密油开采技术发展策略思考[J]. 石油钻探技术, 2015,43(1): 1-6.(ZHANG Yinghong, LU Baoping, CHEN Zuo. Thinking on development strategy of continental tight oil production technology in China[J].Petroleum Drilling Technology,2015,43(1): 1-6.(in Chinese))

[7]严翔. 对油页岩勘探现状的评价分析[J]. 智能城市, 2019,5(14): 84-85.(YAN Xiang. Evaluation and analysis of the present situation of oil shale exploration[J].Intelligent City,2019,5(14): 84-85.(in Chinese))

[8]丁述基. 达西及达西定律[J]. 水文地质工程地质, 1986(3): 33-35.(DING Shuji. Darcy and Darcy’s law[J].Hydrogeology and Engineering Geology,1986(3): 33-35.(in Chinese))

[9]赵永富, 田恩龙, 张国栋. 达西定律与渗流控制[J]. 黑龙江水利科技, 2008,36(4): 65.(ZHAO Yongfu, TIAN Enlong, ZHANG Guodong. Darcy’s law and seepage control[J].Heilongjiang Science and Technology of Water Conservancy,2008,36(4): 65.(in Chinese))

[10]牛玉龙, 王媛, 于可, 等. 裂隙网络非达西渗流REV及非达西系数张量研究[J]. 水利学报, 2020,51(4): 468-478.(NIU Yulong, WANG Yuan, YU Ke, et al. Non-Darcy seepage REV and non-Darcy coefficient tensor in fracture network[J].Journal of Hydraulic Engineering,2020,51(4): 468-478.(in Chinese))

[11]王高峰, 雷友忠, 谭俊领, 等. 低渗透油藏气驱注采比和注气量设计[J]. 油气地质与采收率, 2020,27(1):134-139.(WANG Gaofeng, LEI Youzhong, TAN Junling, et al. Design of injection-production ratio and gas injection rate of gas flooding in low-permeability reservoirs[J].Petroleum Geology and Recovery Efficiency,2020,27(1): 134-139.(in Chinese))

[12]袁银春, 李闽, 王颖, 等. 致密砂岩低流速下的达西流动[J]. 新疆石油地质, 2020,41(3): 349-354.(YUAN Yinchun, LI Min, WANG Ying, et al. Darcy flow in tight sandstone at low velocity[J].Xinjiang Petroleum Geology,2020,41(3): 349-354.(in Chinese))

[13]HUBBERT M K. Energy from fossil fuels[J].Science,1949,109(2823): 103-109.

[14]HUBBERT M K. The energy resources of the earth[J].Scientific American,1971,225(3): 60-70.

[15]WENG W B.Theory of Forecasting[M]. Beijing: International Academic Publishers, 1991.

[16]陈劲松, 年静波, 韩洪宝. 改进Arps递减模型早期产量预测再认识[J]. 非常规油气, 2019,6(1): 75-80.(CHEN Jinsong, NIAN Jingbo, HAN Hongbao. Improve production prediction rationality of the modified Arps decline model in the early shale oil and gas wells[J].Unconventional Oil and Gas,2019,6(1): 75-80.(in Chinese))

[17]REBERTSON S. Generalized hyperbolic equation： SPE-18731-MS[R]. Society of Petroleum Engineers, 1988.

[18]DILHAN I. Exponential vs hyperbolic decline in tight gas sands understanding the origin and implication for reserve estimates using Arps decline curves[C]//SPE Annual Technical Conference and Exhibition. Denver, Colorado, USA, 2008: SPE-116731-MS.

[19]DUONG A N. An unconventional rate decline approach for tight and fracture-dominated gas wells[C]//Canadian Unconventional Resources and International Petroleum Conference. Calgary, Alberta, Canada, 2011: SPE-137748-MS.

[20]陈海虹, 黄彪, 刘峰, 等. 机器学习原理及应用[M]. 成都: 电子科技大学出版社, 2017.(CHEN Haihong, HUANG Biao, LIU Feng, et al.Principle and Application of Machine Learning[M]. Chengdu: University of Electronic Science and Technology Press, 2017.(in Chinese))

[21]ZEILER M D, KRISHNAN D, TATLOR G W, et al. Deconvolutional networks[C]//2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. San Francisco, CA, USA, 2010.

[22]AHRIMANKOSH M, KASIRI N, MOUSAVI S M. Improved permeability prediction of a heterogeneous carbonate reservoir using artificial neural networks based on the flow zone index approach[J].Petroleum Science and Technology,2011,29(23): 2494-2506.

[23]KHANAL A, KHOSHGHADAM M, LEE W J. New forecasting method for liquid rich shale gas condensate reservoirs with data driven approach using principal component analysis[J].Journal of Natural Gas Science and Engineering,2017,38: 621-637.

[24]CAO Q, BANERJEE R, GUPTA S. Data driven production forecasting using machine learning[C]//SPE Argentina Exploration and Production of Unconventional Resources Symposium. Buenos Aires, Argentina, 2016: SPE 180984-MS.

[25]郭长杰, 王浩翔, 刘晓，等. 浅析机器学习技术在油气行业的应用场景[J]. 信息系统工程, 2017(5): 100-103.(GUO Changjie, WANG Haoxiang, LIU Xiao, et al. Analysis of the application scenarios of machine learning technique in the oil and gas industry[J].Information System Engineering,2017(5): 100-103.(in Chinese))

[26]马林茂, 李德富, 郭海湘. 基于遗传算法优化BP神经网络在原油产量预测中的应用:以大庆油田BED试验区为例[J]. 数学的实践与认识, 2015,45(24): 117-128.(MA Linmao, LI Defu, GUO Haixiang. BP neural network based on genetic algorithm applied in crude oil production forecast: taking the BED test area of the Daqing oilfield as an example[J].Mathematics in Practice and Theory,2015,45(24): 117-128.(in Chinese))

[27]檀朝东, BANGERT P, 刘柏良, 等. 人工神经网络自学习方法在大港滩海油田的应用[J]. 中国石油和化工, 2010(11): 46-47.(TAN Chaodong, BANGERT P, LIU Boliang, et al. Application of self-learning method of artificial neural network in Dagang Beach oilfield[J].China Petroleum and Chemical Industry,2010(11): 46-47.(in Chinese))

[28]DAVID E R, GEOFFREY E H, RONALD J W. Learning representations by back-propagating errors[J].Nature,1986,323: 533-536.

[29]袁冰清, 程功, 郑柳刚. BP神经网络基本原理[J]. 数字通信世界, 2018(8): 28-29.(YUAN Bingqing, CHENG Gong, ZHENG Liugang. Basic principle of BP neural network[J].Digital Communication World,2018(8): 28-29.(in Chinese))

[30]周志祥, 韩逢庆. 一种基于训练数据的迭代改进核函数[J]. 应用数学和力学, 2009,30(1): 120-126.(ZHOU Zhixiang, HAN Fengqing. An iterative improved kernel function based on training data[J].Applied Mathematics and Mechanics,2009,30(1): 120-126.(in Chinese))

[31]佟秀秀, 康志宏. 基于多元线性回归和BP神经网络的单井能力预测[J]. 科学技术与工程, 2019,19(29): 96-102.(TONG Xiuxiu, KANG Zhihong. Single well capacity prediction based on multiple linear regression and back propagation neural network[J].Science Technology and Engineering,2019,19(29): 96-102.(in Chinese))

[32]ELMAN J L. Finding structure in time[J].Cognitive Science,1990,149(2): 179-211.

[33]HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J].Neural Compute,1997,9(8): 1735.

[34]张巧灵, 高淑萍, 何迪. 基于时间序列的混合神经网络数据融合算法[J]. 应用数学和力学, 2021,42(1): 12-16.(ZHANG Qiaoling, GAO Shuping, HE Di. A hybrid neural network data fusion algorithm based on time series[J].Applied Mathematics and Mechanics,2021,42(1):12-16.(in Chinese))

[35]陈劲松, 曹健志, 韩洪宝. 页岩油气井常用产量预测模型适应性分析[J]. 非常规油气, 2019,6(3): 48-57.(CHEN Jinsong, CAO Jianzhi, HAN Hongbao. Adaptability analysis of common production prediction models for shale oil and gas wells[J].Unconventional Oil and Gas,2019,6(3): 48-57.(in Chinese))

[36]翁文波. 预测论基础[M]. 北京: 石油工业出版社, 1984.(WENG Wenbo.Fundamentals of Prediction Theory[M]. Beijing: Petroleum Industry Press, 1984.(in Chinese))

[37]王昕, 程希明. 石油产量预测的麦克斯韦模型[J]. 岩性油气藏, 2019,31(6): 155-160.(WANG Xin, CHENG Ximing. Maxwell model for oil production prediction[J].Lithologic Reservoir,2019,31(6): 155-160.(in Chinese))

[38]李彦尊, 白玉湖, 陈桂华. 基于人工神经网络方法的页岩油气产量预测新技术: 以美国Eagle Ford页岩油气田为例[J]. 中国海上油气, 2020,32(4): 104-110.(LI Yanzun, BAI Yuhu, CHEN Guihua. ANN method based on novel technology for production prediction of shale oil and gas: a case study in Eagle Ford[J].China Offshore Oil and Gas,2020,32(4): 104-110.(in Chinese))

[39]李世臻, 刘卫彬, 王丹丹, 等. 中美陆相页岩油地质条件对比[J]. 地质论评, 2017,63（S1）: 39-40.(LI Shizhen, LIU Weibin, WANG Dandan, et al. Comparison of geological conditions of continental shale oil between China and the United States[J].Geological Review,2017,63（S1）: 39-40.(in Chinese))

[40]祝彦贺. 北美某盆地Z区块页岩油气产量的影响因素[J]. 海洋地质前沿, 2013,29（8）: 33-38, 52.(ZHU Yanhe. Study of geologic parameters controlling production difference in Z block of a basin in North America[J].Marine Geology Frontier,2013,29(8): 33-38, 52.(in Chinese))

[41]王民, 石蕾, 王文广, 等. 中美页岩油、致密油发育的地球化学特征对比[J]. 岩性油气藏, 2014,26(3): 67-73.(WANG Min, SHI Lei, WANG Wenguang, et al. Comparative study on geochemical characteristics of shale oil between China and USA[J].Lithologic Reservoirs,2014,26(3): 67-73.(in Chinese))

[42]白国平, 邱海华, 邓舟舟. 美国页岩油资源分布特征与主控因素研究[J]. 石油实验地质, 2020,42(4): 524-532.(BAI Guoping, QIU Haihua, DENG Zhouzhou. Distribution and main controls for shale oil resources in USA[J].Petroleum Geology and Experiment,2020,42(4): 524-532.(in Chinese))

施引文献

资源附件(0)

访问统计

计量

文章访问数: 1206
HTML全文浏览量: 210
PDF下载量: 155
被引次数: 0

姓名
邮箱
手机号码
标题
留言内容
验证码

留言板

基于物理经验模型约束的机器学习方法在页岩油产量预测中的应用

doi: 10.21656/1000-0887.420015

作者简介:
周济民(2000—)，男(E-mail: zhoujm18@mails.tsinghua.edu.cn);王沫然(1977—)，男，教授，博士，博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).

通讯作者:
王沫然(1977—)，男，教授，博士，博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).

计量

Machine Learning With Physical Empirical Model Constraints for Prediction of Shale Oil Production

计量

目录

留言板

基于物理经验模型约束的机器学习方法在页岩油产量预测中的应用

doi: 10.21656/1000-0887.420015

作者简介: 周济民(2000—)， 男(E-mail: zhoujm18@mails.tsinghua.edu.cn);王沫然(1977—)， 男， 教授， 博士， 博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).

通讯作者: 王沫然(1977—)， 男， 教授， 博士， 博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).

计量

出版历程

Machine Learning With Physical Empirical Model Constraints for Prediction of Shale Oil Production

计量

出版历程

目录

作者简介:
周济民(2000—)，男(E-mail: zhoujm18@mails.tsinghua.edu.cn);王沫然(1977—)，男，教授，博士，博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).

通讯作者:
王沫然(1977—)，男，教授，博士，博士生导师(通讯作者. E-mail: mrwang@tsinghua.edu.cn).