BP神经网络与多点地质统计相结合的井震约束浊积水道模拟
|
摘要
勘探阶段,仅依靠数目较少的井点数据很难在较大工区范围内建立令人信服的沉积相模式。然而,以地震数据做约束,往往会使沉积相更加逼近软数据的分布趋势。以澳西北陆架Vulcan次盆P22区块Puffin组Unit7小层为研究对象,在建立沉积相的过程中,提出了分层次约束的方法,即以BP神经网络算法控宏观,以多点地质统计学约微观的方法流程,同时结合已知井的岩性类型数据点和源于地震的岩性类型数据点,再现了浊积水道的沉积展布规律。采用截断高斯和多点地质统计学2种方法进行模拟,分别产生了浊积水道的8个模拟实现,对比结果表明,多点地质统计学模拟方法在再现浊积水道的几何形态方面比基于变差函数的截断高斯模拟更具有优越性。最后以偏差目标函数为标准,优选出了偏差最小的模拟实现。
At oil exploriation stage,it was hard to build the convinced sedimentary facies modes with less well-point data in the large operation area.However,for the restriction of seismic data,the sedimentary facies was often more approximate to the distributive tendency of soft data.In the process of building the sedimentary facies,Subzone 7 of Puffin Formation in Block P22 in the Vulcan Sub-basin of north-west Australia continent shelf was used as studied object,a layering constraining method was proposed.The idea that macroscale of modeling was controlled by BP arithmetic and micro-shape of geological body was depicted by multiple-point geostatistics propounded in the modeling progress,and in combination with the lithologic type data derived from known well data and seismic data,the distribution of turbidite channel could be realized by modeling.Truncated Gaussian simulation and multiple-point geostatistics are applied in the modeling process,and each method responds to 8 simulation processes respectively.The comparison result indicates that multiple-point geostatistics is more proper than variogram based Truncated Gaussian simulation in the process of characterizing the geometry of turbidite channel.The optimal simulation with less errors is obtained by the criterion of minimal value of bias object function.
At oil exploriation stage,it was hard to build the convinced sedimentary facies modes with less well-point data in the large operation area.However,for the restriction of seismic data,the sedimentary facies was often more approximate to the distributive tendency of soft data.In the process of building the sedimentary facies,Subzone 7 of Puffin Formation in Block P22 in the Vulcan Sub-basin of north-west Australia continent shelf was used as studied object,a layering constraining method was proposed.The idea that macroscale of modeling was controlled by BP arithmetic and micro-shape of geological body was depicted by multiple-point geostatistics propounded in the modeling progress,and in combination with the lithologic type data derived from known well data and seismic data,the distribution of turbidite channel could be realized by modeling.Truncated Gaussian simulation and multiple-point geostatistics are applied in the modeling process,and each method responds to 8 simulation processes respectively.The comparison result indicates that multiple-point geostatistics is more proper than variogram based Truncated Gaussian simulation in the process of characterizing the geometry of turbidite channel.The optimal simulation with less errors is obtained by the criterion of minimal value of bias object function.
引文
[1]Rumelhart D E,Hinton G E,Williams R J.Learning representations by back-propagating errors[J].Nature,1986,323(9):533～536.
[2]Kamel B,Noureddine D,Tahar A,et al.Acoustic impedance inversion by feedback artificial neural network[J].Journal of PetroleumScience and Engineering,2010,71(3-4):106～111.
[3]Noureddine D,Tahar A,Karnel B,et al.Application of feedback connection artificial neural network to seismic data filtering[J].Comptes Rendus Geoscience,2008,340(6):335～344.
[4]Helle H B,Bhatt A,Ursin B.Porosity and permeability prediction from wireline logs using artificial neural networks:a North Sea casestudy[J].Geophysical Prospecting,2001,49(4):431～444.
[5]Calderon-Macias C,Sen M K,Stoffa P L.Artificial neural networks for parameter estimation in geophysics[J].Geophysical Prospec-ting,2000,48(1):21～47.
[6]Guardiano F B,Srivastava R M.Multivariate geostatistics——Beyond bivariate moments[A].Soares A.Geostatistics-Troia[C].Dordrecht;Boston;London:Kluwer Academic Publications,1993.133～144.
[7]Strebelle S.Multiple-point statistics simulation of continuous variables[J].GIS and Spatial Analysis,2005,(1):732～736.
[8]Strebelle S,Journel A.Sequential simulation drawing structures from training images[Z].Stanford Center for Reservoir ForecastingAnnual Report,2000.381～392.
[9]于兴河.油气储层表征与随机建模的发展历程及展望[J].地学前缘,2008,15(1):1～15.
[10]Yu Xinghe,Li Shengli,Zhao Shu,et al.Constraining method of stochastic modeling for fluvial petroleum reservoir controlled by depo-sitional facies using wells and seismic data[J].Earth Science Frontiers,2008,15(4):25～33.
[11]蒋宗礼.人工神经网络导论[M].北京:高等教育出版社,2001.18～19.
[12]王彦春,苑春芳.静校正及神经网络处理技术[M].北京:地质出版社,2000.74～75.
[13]Nichols G.Sedimentology and Stratigraphy[M].Second Edition.UK:John Wiley&Sons Ltd,2009.255～256.
[2]Kamel B,Noureddine D,Tahar A,et al.Acoustic impedance inversion by feedback artificial neural network[J].Journal of PetroleumScience and Engineering,2010,71(3-4):106～111.
[3]Noureddine D,Tahar A,Karnel B,et al.Application of feedback connection artificial neural network to seismic data filtering[J].Comptes Rendus Geoscience,2008,340(6):335～344.
[4]Helle H B,Bhatt A,Ursin B.Porosity and permeability prediction from wireline logs using artificial neural networks:a North Sea casestudy[J].Geophysical Prospecting,2001,49(4):431～444.
[5]Calderon-Macias C,Sen M K,Stoffa P L.Artificial neural networks for parameter estimation in geophysics[J].Geophysical Prospec-ting,2000,48(1):21～47.
[6]Guardiano F B,Srivastava R M.Multivariate geostatistics——Beyond bivariate moments[A].Soares A.Geostatistics-Troia[C].Dordrecht;Boston;London:Kluwer Academic Publications,1993.133～144.
[7]Strebelle S.Multiple-point statistics simulation of continuous variables[J].GIS and Spatial Analysis,2005,(1):732～736.
[8]Strebelle S,Journel A.Sequential simulation drawing structures from training images[Z].Stanford Center for Reservoir ForecastingAnnual Report,2000.381～392.
[9]于兴河.油气储层表征与随机建模的发展历程及展望[J].地学前缘,2008,15(1):1～15.
[10]Yu Xinghe,Li Shengli,Zhao Shu,et al.Constraining method of stochastic modeling for fluvial petroleum reservoir controlled by depo-sitional facies using wells and seismic data[J].Earth Science Frontiers,2008,15(4):25～33.
[11]蒋宗礼.人工神经网络导论[M].北京:高等教育出版社,2001.18～19.
[12]王彦春,苑春芳.静校正及神经网络处理技术[M].北京:地质出版社,2000.74～75.
[13]Nichols G.Sedimentology and Stratigraphy[M].Second Edition.UK:John Wiley&Sons Ltd,2009.255～256.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心