摘要
介绍深度学习的基本理论方法及基于地震数据深度学习的深层天然气储层预测。指出了深层油气储层的"三弱"特点和预测所涉及的问题与面临的困难;提出了深度学习的深网络构建与训练方法,基于地震数据深度学习的深层天然气储层预测进行实验。认为深度学习(Deep learning)揭示对象内秉特征的能力,有助于发现使用现有方法难以发现的油气储层地震响应的内禀特征。同时探讨了基于地震数据深度学习的油气储层预测所涉及的关键性科学技术问题。
This paper introduces deep learning(DL)and deep gas reservoirs prediction based on the deep learning of seismic data.The paper introduces the "three weak" characteristics of deep oil and gas reservoirs and the key problems have to be dealt with in the reservoir recognition.The paper introduces the construction and training methods of deep network in deep learning,and an experimental result of deep gas reservoirs prediction based on the deep learning of seismic data.The experimental result shows that deep learning is of ability to reveal the intrinsic features of the seismic responses of hydrocarbon reservoirs.The paper also discusses the key scientific and technological issues involved in the prediction of deep oil and gas reservoirs using deep learning.
引文
[1]《BP世界能源统计年鉴》2017版[EB/OL].[2017-07-05].https://www.bp.com/zh_cn/china/reportsand-publications/_bp_2017-_.html.BP world energy statistics yearbook 2017Edition[EB/OL].[2017-07-05].https://www.bp.com/zh_cn/china/reports-and-publications/_bp_2017-_.html.(In Chinese)
[2]国家发展改革委,国家能源局.能源生产和消费革命战略(2016-2030)[EB/OL].[2017-04-25].http://www.gov.cn/xinwen/2017-04/25/5230568/files/286514af354e41578c57ca38d5c4935b.pdf.National development and Reform Commission,National Energy Agency,"energy production and consumption revolution strategy(2016-2030)"[EB/OL].[2017-04-25].http://www.gov.cn/xinwen/2017-04/25/5230568/files/286514af354e41578c57ca38d5c4935b.pdf.(In Chinese)
[3]中国石油第四次油气资源评价重大科技专项收官[EB/OL].[2016-11-11]http://www.mzxww.net/syzy/201612/387.html.Oil and gas resources assessment of China's fourth major oil and gas resources[EB/OL].[2016-11-11].http://www.mzxww.net/syzy/201612/387.html.(In Chinese)
[4]刘树根,孙玮,赵异华,等.四川盆地震旦系灯影组天然气的差异聚集分布及其主控因素[J].天然气工业,2015,35(1):10-23.LIU S G,SUN W,ZHAO Y H,et al.Differential accumulation and distribution of natural gas and their main controlling factors in the Upper Sinian Dengying Fm,Sichuan Basin[J].Natural Gas Industry,2015,35(1):10-23.(In Chinese)
[5]HAMMOND A.L.Bright Spot:Better seismological indicators of gas and oil[J].Science,1974,185:515-517.
[6]OSTRANDER W.J.Plane wave reflection coefficients for gas sands at non-normal angles of incidence[J].Geophysics,1984,49(10):1637-1648.
[7]SHUEY R.T.A simplification of the Zoeppritz equation[J].Geophysics,1985,50(4):609-614.
[8]EBROM D.2004.The low-frequency gas shadow on seismic sections[J].The Leading Edge,2004,23(8):772-772.
[9]BACLIS,M.M.,CHEN,L.Flat-spot exploration[J].Geophysical Prospecting.1975,23(3):533-577.
[10]BROWN,A.R.Dim spots:Opportunity for future hydrocarbon discoveries[J].The Leading Edge,2012,31(6),682-683.
[11]JUNXING CAO,XIAOYAN HE,RENFEI TIAN.Seismic-print analysis and hydrocarbon identification[C].AGU FALL Meeting 2011.
[12]曹俊兴、田仁飞.基于地震纹分析的储层检测方法.中国.ZL 2011 1 0165615.6[P].2013.CAO J X,TIAN R F.Reservoir detection method based on seismic-print analysis[P].Patent No.:ZL2011 1 0165615.6.2013.
[13]曹俊兴,刘树根,田仁飞,等.龙门山前陆盆地深层海相碳酸盐岩储层地震预测研究[J].岩石学报,2011,27(8):2423-2434.CAO J X,LIU S G,TIAN R F,et al.Seismic prediction of carbonate reservoirs in the deep of Longmenshan foreland basin[J].Acta Petrologica Sinica,2011,27(8):2423-2434.(In Chinese)
[14]AlphaGo Zero[DB/OL].[2017-11-17].https://en.wikipedia.org/wiki/AlphaGo_Zero.
[15]Deep_learning[DB/OL].[2017-11-23]..https://en.wikipedia.org/wiki/Deep_learning.
[16]曹俊兴.基于地震纹特征深度学习的天然气储层识别[C].2016中国地球科学联合学术年会论文集.2016.CAO J X.Gas reservoir identification basing on deep learning of seismic-print characteristics[C].Proceedings of the 2016China Earth Science Conference.2016.(In Chinese)
[17]CAO J.,WU S.,He X.Gas reservoir identification basing on deep learning of seismic-print Characteristics[C].AGU Fall Meeting Abstracts.2016.
[18]CAO J X,WU Shikai.Deep Learning:Chance and challenge for deep gas reservoir identification[C].International Geophysical Conference,Qingdao,China.2017.
[19]吴施楷.基于深度学习的天然气储层检测方法研究[D].成都:成都理工大学,2017.WU S K.Research on gas reservoir identification method based on deep learning[D].Chengdu:Chengdu University of Technology,2017.(In Chinese)
[20]KEITH HOLDAWAY,PATRICK HALL,JORGE SILVA,et al.Surface reservoir characteristics from subsurface seismic images with deep learning methodologies[C].International Conference and Exhibition,Barcelona,Spain,3-6April 2016.Society of Exploration Geophysicists and American Association of Petroleum Geologists,2016.
[21]KORJANI M.,POPA A.,GRIJALVA E.,et al.A new approach to reservoir characterization using deep learning neural networks[C].SPE Western Regional Meeting,Anchorage,Alaska USA.Society of Petroleum Engineers,2016.
[22]LU LIU ZHANG GUANGZHI,ZHAO CHEN.Reservoir thickness forecasting based on Deep Belief Networks[C].International Geophysical Conference,Qingdao,China.2017.https://doi.org/10.1190/IGC2017-186.
[23]MAURICIO ARAYA-POLO,TAYLOR DAHLKE,CHARLIE FROGNER,et al.Automated fault detection without seismic processing[J].The Leading Edge,2017,36(3),208-214.
[24]XU,P.C.,LU,WENKAI,WANG,BENFENG.Multi-Attribute Classification Based On Sparse Autoencoder-A Gas Chimney Detection Example[C].79th EAGE Conference and Exhibition,Paris,France.2017.DOI:10.3997/2214-4609.201700920.
[25]GOODFELLOW I.,BENGIO Y.,COURVILLE A.Deep Learning[M],Cambridge,Massachusetts:MIT Press.2016.
[26]ROSENBLATT F.The Perceptron:A probabilistic model for information storage and organization in the brain[J].Cornell Aeronautical Laboratory,Psychological Review,1958,65,386-408.
[27]IVAKHNENKO,A.G.,LAPA,V.G..Cybernetic predicting devices[M].CCM Information Corporation.1965.
[28]HINTON,G.E.,SALAKHUTDINOV,R.Reducing the dimensionality of data with Neural Networks[J].Science,2006,313(5786):504-507.
[29]HINTON G.,OSINDERO S.,TEH Y.-W.A fast learning algorithm for deep belief nets[J].Neural Computation,2006,18:1527-1554.
[30]WILLIAMS D,HINTON G.Learning representations by back-propagating errors[J].Nature,1986,323(6088):533-538.
[31]SCHMIDHUBER J.Deep learning in neural networks:An overview[J].Neural networks,2015,61:85-117.
[32]SABOUR S.,FROSST N.,HINTON G E.Dynamic routing between capsules[DB/OL].[2017-10-26].arXiv preprint arXiv:1710.09829,2017.
[33]SHI S,WANG Q,XU P,et al.Benchmarking state-of-the-art deep learning software tools[DB/OL].[2016-08-25].arXiv preprint arXiv:1608.07249.
