LBFGS CSAMT 1D inversion of limited memory based on boundary constraint and its application

Authors

TANG Chuanzhang, Huabei Oil Reservoir, CNPC, Renqiu 062552, China
CHENG Jianzhong, Key Laboratory of Exploration Technologies for Oil and Gas Resources, Yangtze University, Wuhan 430100, China
YAN Liangjun, Key Laboratory of Exploration Technologies for Oil and Gas Resources, Yangtze University, Wuhan 430100, ChinaFollow
FENG Guangye, Huabei Oil Reservoir, CNPC, Renqiu 062552, China
ZHOU Lei, Key Laboratory of Exploration Technologies for Oil and Gas Resources, Yangtze University, Wuhan 430100, China
XU Fengjiao, Key Laboratory of Exploration Technologies for Oil and Gas Resources, Yangtze University, Wuhan 430100, China

Abstract

CSAMT is characterized by time-consuming calculation and low accuracy. The conventional inversion method has simple boundary processing, needs big memory, it is difficult to have accurate search direction, influencing severely the effect of inversion. In order to achieve rapid and high-precision forward modeling of CSAMT, to improve the boundary processing method, to reduce memory usage, to optimize the searching direction to reduce the times of iterative inversion, LBFGS CSAMT 1D inversion method based on boundary constraint and limited memory was proposed. Firstly in CSAMT forward modeling, double-precision and parallel algorithm was adopted, realizing rapid and high-precision calculation of the horizontal electric field excited by a wire source with limited length under conditions of 1D medium model. Secondly in inversion method two objective functions based on the relative error and the absolute error were constructed respectively. The constraints of smooth model were introduced. Self-adaptive regularization strategy was adopted to renewal the regularization factors. LBFGS-B algorithm of limited memory based on boundary constraint was adopted. 1D precise and rapid inversion of CSAMT was realized. The amplitude of the horizontal electric field E_x of the geoelectric model of multiple thin layers with low and high resistivity was used as inversion data to conduct direct inversion and model test. The results illustrate that this method can reflect the trend of the variation of formation resistivity with depth, and has higher resolution for thin layers with high and low resistivity, the resolving effect is better for a thin layer with low resistivity. The resolution of the inversion profile of the actual CSAMT data is superior to that of the conventional inversion method for continuous medium. The geological effect is evident, indicating that this method has good application prospect.

Keywords

CSAMT forward modeling, parallel algorithm, objective function, LBFGS-B inversion

DOI

10.3969/j.issn.1001-1986.2019.05.027

Recommended Citation

TANG Chuanzhang, CHENG Jianzhong, YAN Liangjun, et al. (2019) "LBFGS CSAMT 1D inversion of limited memory based on boundary constraint and its application," Coal Geology & Exploration: Vol. 47: Iss. 5, Article 28.
DOI: 10.3969/j.issn.1001-1986.2019.05.027
Available at: https://cge.researchcommons.org/journal/vol47/iss5/28

Reference

[1] LU Xinyu,MARTYN U,JOHN B. Rapid relaxation inver-sion of CSAMT data[J]. Geophysical Journal of the Royal Astronomical Society,1999,138(2):381-392.

[2] 张凯飞. 基于电性约束的NLCG反演在CSAMT资料中的应用[J]. 物探与化探,2016,40(3):629-634. ZHANG Kaifei. The application of CSAMT data processed with NLCG inversion based on electrical constraints[J]. Geophysical and Geochemical Exploration,2016,40(3):629-634.

[3] 陈明生. 关于频率电磁测深几个问题的探讨(一):从可控源音频大地电磁测深原理看解释中的问题[J]. 煤田地质与勘探,2012,40(5):63-66. CHEN Mingsheng. Problem of data interpretation from the view of principle of controlled-source audio-frequency magnetotellurics[J]. Coal Geology & Exploration,2012,40(5):63-66.

[4] 严良俊,胡文宝,杨绍芳,等. 电磁勘探方法及其在南方碳酸盐岩地区的应用[M]. 北京:石油工业出版社,2001.

[5] ROUTH P S,OLDENBURG D W. Inversion of controlled source audio-frequency magnetotellurics data for a horizontally layered earth[J]. Geophysics,1999,64(6):1689-1697.

[6] 王若,王妙月. 一维全资料CSAMT反演[J]. 石油地球物理勘探,2007,42(1):107-114. WANG Ruo,WANG Miaoyue. One-dimensional full-data CSAMT inversion[J]. Oil Geophysical Prospecting,2007,42(1):107-114.

[7] 蒋齐平. 基于场值数据的可控源音频大地电磁一维全资料反演研究[J]. 科学技术与工程,2015,15(3):193-196. JIANG Qiping. Research on 1D full CSAMT data based on filed data[J]. Science Technology and Engineering,2015,15(3):193-196.

[8] 尚通晓,李桐林,关艺晓,等. CSAMT一维全区反演[J]. 吉林大学学报(地球科学版),2007,37(增刊1):29-31. SHANG Tongxiao,LI Tonglin,GUAN Yixiao,et al. Full-region inversion of 1-D CSAMT[J]. Journal of Jilin University(Earth Science Edition),2007,37(S1):29-31.

[9] 朱威,李桐林,尚通晓,等. CSAMT一维全区反演对比研究[J]. 吉林大学学报(地球科学版),2008,38(增刊1):12-14. ZHU Wei,LI Tonglin,SHANG Tongxiao,et al. Compared study of 1-D full-region inversion of CSAMT[J]. Journal of Jilin University(Earth Science Edition),2008,38(S1):12-14.

[10] 周军,袁伟,连晨龙. 可控源音频大地电磁法(CSAMT)一维全区Levenberg-Marquardt反演方法[J]. 科学技术与工程,2014,14(14):129-134. ZHOU Jun,YUAN Wei,LIAN Chenlong. Leven-berg-Marquard inversion method of CSAMT one-dimension region data[J]. Science Technology and Engineering,2014,14(14):129-134.

[11] 汤井田,张林成,肖晓. 基于频点CSAMT一维最小构造反演[J]. 物探化探计算技术,2011,33(6):577-581. TANG Jingtian,ZHANG lincheng,XIAO Xiao. One-dimensional minimum structure inversion based on frequency CSAMT[J]. Computing Techniques for Geophysical and Geochemical Exploration,2011,33(6):577-581.

[12] 刘颖,李予国,柳建新,等. 海洋可控源电磁场的一维反演[J]. 中国有色金属学报,2013,23(9):2551-2556. LIU Ying,LI Yuguo,LIU Jianxin,et al. One-dimension inversion of marine controlled-source electromagnetic fields[J]. The Chinese Journal of Nonferrous Metal,2013,23(9):2551-2556.

[13] 王堃鹏,曹辉,高妍,等. CSAMT自适应正则化一维全资料反演[J]. 物探化探计算技术,2013,35(5):530-533. WANG Kunpeng,CAO Hui,GAO Yan,et al. Adaptive regularized inversion of 1-D full CSAMT data[J]. Computing Techniques for Geophysical and Geochemical Exploration,2013,35(5):530-533.

[14] 李帝铨,王光杰,底青云,等. 基于遗传算法的CSAMT最小构造反演[J]. 地球物理学报,2008,51(4):1234-1245. LI Diquan,WANG Guangjie,DI Qingyun,et al. The ap-plication of genetic algorithm to CSAMT inversion for min-imum structure[J]. Chinese Journal of Geophysics,2008,51(4):1234-1245.

[15] LIU D C,NOCEDAL J. On the limited memory BFGS method for large scale optimization[J]. Mathematical Programming:Series B,1989,45(3):503-528.

[16] 孙彩堂,李玲,黄维宁,等. 基于自适应遗传算法的CSAMT一维反演[J]. 石油地球物理勘探,2017,52(2):392-397. SUN Caitang,LI ling,HUANG Weining,et al. 1D inversion of CSAMT data based on adaptive genetic algorithm[J]. Oil Geophysical Prospecting,2017,52(2):392-397.

[17] 贾定宇. 基于L-BFGS方法的水平电偶极一维反演[D]. 长春:吉林大学,2012.

[18] 翁爱华,刘佳音,贾定宇,等. 有限长导线源频率测深有限内存拟牛顿一维反演[J]. 吉林大学学报(地球科学版),2017,47(2):597-605. WENG Aihua,LIU Jiayin,JIA Dingyu,et al. 1-D inversion for controlled source electromagnetic sounding using limited memory quasi-Newton method[J]. Journal of Jilin University(Earth Science Edition),2017,47(2):597-605.

[19] AVDEEV D,AVDEEVA A. 3D magnetotelluric inversion using a limited-memory quasi-Newton optimization[J]. Geo-physics,2009,74(3):45-57.

[20] BYRD R H,LU P,NOCEDAL J,et al. A limited memory algorithm for bound constrained optimization[J]. SIAM Journal on Scientific Computing,1995,16(5):1190-1208.

[21] 阮帅. 三维大地电磁有限内存拟牛顿反演[D]. 成都:成都理工大学,2015.

[22] ZHU C,BYRD R H,LU P,et al. LBFGS-B:Fortran subroutines for large-scale bound constrained optimization[J]. Report NAM-11,EECS Department,Northwestern University,1994.

[23] MORALES J L,NOCEDAL J. Remark on "Algorithm 778:L-BFGS-B:Fortran subroutines for large-scale bound con-strained optimization"[J]. ACM Transactions on Mathematical Software(TOMS),2011,38(1):7.

[24] 陈小斌,赵国泽,汤吉,等. 大地电磁自适应正则化反演算法[J]. 地球物理学报,2005,48(4):1005-1016. CHEN Xiaobin,ZHAO Guoze,TANG Ji,et al. The adaptive regularized inversion algorithm for magnetotelluric data[J]. Chinese Journal of Geophysics,2005,48(4):1005-1016.

[25] 严良俊,胡文宝,陈清礼,等. 长偏移距瞬变电磁测深的全区视电阻率求取及快速反演方法[J]. 石油地球物理勘探,1999,34(5):532-538. YAN Liangjun,HU Wenbao,CHEN Qingli,et al. The estimation and fast inversion of all-time apparent resistivities in long-offset transient electromagnetic souding[J]. Oil Geophysical Prospecting,1999,34(5):532-538.