Gravity inversion method based on quasi-neural network featuring Gaussian radial basis function
XIANG Peng1, TAN Shaoquan1, CHEN Xueguo1, LIU Jia2
1. Research Institute of Exploration & Development of Shengli Oilfield, SINOPEC, Dongying, Shandong 257000, China; 2. Petroleum Development Center of Shengli Oilfield, SINOPEC, Dongying, Shandong 257000, China
Abstract：An inversion method based on a quasi-neural network featuring Gaussian radial basis function (RBF) is presented in this paper to improve the resolution of gravity inversion. The model space is compressed through the Gaussian RBF, and the dimension of the inversion parameters is reduced without influencing the representation ability of complex models. A quasi-neural network structure is proposed which takes the Gaussian RBF as the activation function and saves the difficulty of establishing a training set in that it does not require training. The proposed method solves the pro-blems of skinning, low vertical resolution, strong multi-solution, and severe dependence on priori constraints caused by the ill-posedness of gravity inversion. In addition, it can extract effective information from gravitational data to enhance the resolution and reliability of the inversion results. Model experiments show that the method, with high accuracy and resolution, can accurately obtain the position, boundary, and density of the model through inversion. A residual density model with a high vertical resolution is obtained by inverting the gravitational data of the Chezhen Depression. Density interfaces and profiles are extracted from the density model for structural interpretation, which in turn reveals the structural pattern of Lower Paleozoic and the development law of buried hills and proves the practical value and application potential of this method.
张永明, 张贵宾.二维重力"等维"反演研究[J]. 工程地球物理学报, 2009, 6(2):137-142.ZHANG Yongming, ZHANG Guibin.Research on the equivalent dimensions of 2-D gravity[J]. Chinese Journal of Engineering Geophysics, 2009, 6(2):137-142.
[2]
王逸宸.重力等维聚焦反演研究[D].北京:中国地质大学(北京), 2014.WANG Yichen.Study of Equidimension Focusing Inversion of Gravity Data[D].China University of Geosciences(Beijing), Beijing, 2014.
[3]
王浩然, 陈超, 杜劲松.重力梯度张量数据的三维反演方法与应用[J]. 石油地球物理勘探, 2013, 48(3):474-481.WANG Haoran, CHEN Chao, DU Jinsong.3-D inversion of gravity gradient tensor data and its application[J]. Oil Geophysical Prospecting, 2013, 48(3):474-481.
[4]
耿美霞.基于地质统计学的重力梯度全张量数据三维反演方法研究[D].吉林长春:吉林大学, 2015.GENG Meixia.The Study of 3D Inversion of Full Tensor Gravity Gradiometry Data Based on Geostatistics Method[D].Jilin University, Changchun, Jilin, 2015.
[5]
Sun J J, Li Y G.Inversion of surface and borehole gra-vity with thresholding and density constraints[C].SEG Technical Program Expanded Abstracts, 2010, 29:1798-1803.
[6]
秦朋波, 黄大年.重力和重力梯度数据联合聚焦反演方法研究[J]. 地球物理学报, 2016, 59(6):2203-2224.QIN Pengbo, HUANG Danian.Integrated gravity and gravity gradient data focusing inversion[J]. Chinese Journal of Geophysics, 2016, 59(6):2203-2224.
[7]
王泰涵, 马国庆, 熊盛青, 等.空-地-井重力异常正则化协同密度反演方法[J]. 地球物理学报, 2020, 63(7):2737-2750.WANG Taihan, MA Guoqing, XIONG Shengqing, et al.Joint regularized density inversion method of airborne, surface and borehole gravity anomaly data[J]. Chinese Journal of Geophysics, 2020, 63(7):2737-2750.
[8]
Li Y G, Oldenburg D W.3-D inversion of magnetic data[J]. Geophysics, 1996, 61(2):394-408.
[9]
Portniaguine O, Zhdanov M S.3-D magnetic inversion with data compression and image focusing[J]. Geophysics, 2002, 67(5):1532-1541.
[10]
Zhdanov M S.Geophysical Inverse Theory and Regularization Problems[M].Elsevier Science, Salt Lake City, 2002.
[11]
彭国民, 刘展, 于会臻, 等.基于柯西分布约束和快速近端目标函数优化的三维重力反演方法[J]. 地球物理学报, 2018, 61(12):4934-4941.PENG Guomin, LIU Zhan, YU Huizhen, et al.3D gravity inversion based on Cauchy distribution constraint and fast proximal objective function optimization[J]. Chinese Journal of Geophysics, 2018, 61(12):4934-4941.
[12]
谷文彬, 陈清礼, 王余泉, 等.饶阳凹陷虎8北潜山重力三维松约束反演[J]. 石油地球物理勘探, 2016, 51(6):1219-1226.GU Wenbin, CHEN Qingli, WANG Yuquan, et al.Part-constrained 3D gravity inversion for the Hubabei buried hill in Raoyang Sag[J]. Oil Geophysical Prospecting, 2016, 51(6):1219-1226.
[13]
李海龙, 吴招才, 纪飞, 等.南海北部地壳密度结构:基于约束三维重力反演[J]. 地球物理学报, 2020, 63(5):1894-1912.LI Hailong, WU Zhaocai, JI Fei, et al.Crustal density structure of the South China Sea from constrained 3-D gravity inversion[J]. Chinese Journal of Geophysics, 2020, 63(5):1894-1912.
[14]
Ren Z Y, Zhong Y Y, Chen C J, et al.Gravity anomalies of arbitrary 3D polyhedral bodies with horizontal and vertical mass contrasts up to cubic order[J]. Geophysics, 2018, 83(1):G1-G13.
[15]
Wu L.Fourier-domain modeling of gravity effects caused by polyhedral bodies[J]. Journal of Geodesy, 2018, 93(6):1-19.
[16]
周印明, 戴世坤, 李昆, 等.复杂形体重磁位场三维高效高精度数值模拟[J]. 石油地球物理勘探, 2020, 55(5):1149-1159, 1168.ZHOU Yinming, DAI Shikun, LI Kun, et al.Three-dimensional high-efficiency and high-precision numerical simulation of gravity and magnetic potential fields of complex body[J]. Oil Geophysical Prospecting, 2020, 55(5):1149-1159, 1168.
[17]
刘洁, 张建中, 江丽, 等.基于高阶多项式密度函数的重力反演[J]. 石油地球物理勘探.2019, 54(3):700-708.LIU Jie, ZHANG Jianzhong, JIANG Li, et al.Gravity inversion by high-order polynomial function of density contrast varying with depth[J]. Oil Geophysical Prospecting, 2019, 54(3):700-708.
[18]
Tontini F C, Faggioni O, Beverini N, et al.Gaussian envelope for 3D geomagnetic data inversion[J]. The Leading Edge, 2003, 68(13):996-1007.
[19]
Donoho D L.Compressed sensing[J]. IEEE Transactions on Information Theory, 2006, 52(4):1289-1306.
[20]
Candès E J.The restricted isometry property and its implications for compressed sensing[J]. Comptes Rendus Mathematique, 2008, 346(9-10):589-592.
[21]
Aharon M, Elad M, Bruckstein A.K-SVD:An algorithm for designing overcomplete dictionaries for sparse re-presentation[J]. IEEE Transactions on Signal Proce-ssing, 2006, 54(11):4311-4322.
[22]
Elad M, Aharon M.Image denoising via sparse and redundant representations over learned dictionaries[J]. IEEE Tip, 2006, 15:3736-3745.
[23]
郭文斌, 朱自强, 鲁光银.重力异常的BP神经网络三维物性反演[J]. 地球物理学进展, 2012, 27(2):409-416.GUO Wenbin, ZHU Ziqiang, LU Guangyin.3-D gra-vity inversion for physical properties using BP network[J]. Progress in Geophysics, 2012, 27(2):409-416.
[24]
耿美霞, 杨庆节.应用RBF神经网络反演二维重力密度分布[J]. 石油地球物理勘探, 2013, 48(4):651-657.GENG Meixia, YANG Qingjie.2-D density inversion with RBF neural network method[J]. Oil Geophysical Prospecting, 2013, 48(4):651-657.
[25]
朱自强, 程方道, 黄国祥.同时反演两个三维密度界面的拟神经网BP算法[J]. 石油物探, 1995, 34(1):76-85.ZHU Ziqiang, CHENG Fangdao, HUANG Guoxiang.Quasi neural network BP algorithm for simultaneous inversion of two 3-D density interfaces[J]. Geophysical Prospecting for Petroleum, 1995, 34(1):76-85.
[26]
管志宁, 侯俊胜, 黄临平, 等.重磁异常反演的拟BP神经网络方法及其应用[J]. 地球物理学报, 1998, 41(2):242-251.GUAN Zhining, HOU Junsheng, HUANG Linping, et al.Inversion of gravity and magnetic anomalies using pseudo-BP neural network method and its application[J]. Chinese Journal of Geophysics, 1998, 41(2):242-251.
[27]
唐冲, 惠辉辉.基于Matlab的高斯曲线拟合求解[J]. 计算机与数字工程, 2013, 41(8):1262-1263.TANG Chong, HUI Huihui.Gaussian curve fitting solution based on Matlab[J]. Computer & Digital Engineering, 2013, 41(8):1262-1263.
[28]
Li X, Chouteau M.Three-dimensional gravity mode-ling in all space[J]. Surveys in Geophysics, 1998, 19(4):339-368.
[29]
Tropp J A, Gilbert A C.Signal recovery from random measurements via orthogonal matching pursuit[J]. IEEE Transactions on Information Theory, 2007, 53(12):4655-4666.
[30]
Lin J, Li S.Sparse recovery with coherent tight frames via analysis Dantzig selector and analysis LASSO[J]. Applied and Computational Harmonic Analysis, 2013, 37(1):126-139.
[31]
Engan K, Aase S O, Hakon Husoy J.Method of optimal directions for flame design[C].Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing, Phoenix, USA, 1999, 2443-2446.
[32]
Fernando J S, Valéria C F, João B C.3D gravity inversion through an adaptive-learning procedure[J]. Geophysics, 2009, 74(3):I9-I21.
[33]
胡加山, 胡贤根, 崔世凌.东营凹陷典型剖面重磁电震综合解释方法研究[J]. 石油物探, 2009, 48(4):401-406.HU Jiashan, HU Xian'gen, CUI Shiling.Gravity-magnetic-electric-seismic integrated interpretation method for typical profile of Dongying sag[J]. Geophysical Prospecting for Petroleum, 2009, 48(4):401-406.
