基于深度学习的中心回线瞬变电磁全区视电阻率计算

doi:10.11720/wtyht.2021.1511

摘要
图/表
参考文献
相关文章
Metrics

全文: PDF(1057 KB) HTML
输出: BibTeX | EndNote (RIS)

摘要

深度学习是人工神经网络算法的扩展,对复杂函数有很好的逼近能力,本文将其引入用于瞬变电磁视电阻率计算。首先,建立归一化感应电动势与瞬变场参数单一映射关系的5层深度神经网络,通过对单一隐含层不同神经元个数所训练的误差情况进行分析,确定5层深度神经网络各隐含层神经元个数为13,8,5,8,13。训练算法选择了改进的具有自适应学习率的Nadam算法,该算法可加速训练过程。对训练好的深度神经网络模型进行仿真实验,采用典型地电模型加以验证,发现其对不同的地电模型均具有较好的反映,证明本文采用的基于深度学习计算视电阻率的可行性。应用结果表明训练好的深度神经网络模型可快速准确计算视电阻率。

	服务

	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	吴国培
	张莹莹
	张博文
	赵华亮

关键词 ：瞬变电磁法, 深度学习, 人工神经网络, 视电阻率

Abstract：

Deep learning is an extension of the artificial neural network algorithm, which has a good approximation ability for complex functions. This paper introduces this means for the calculation of transient electromagnetic apparent resistivity. First, a 5-layer deep neural network is established with a single mapping relationship between the normalized induced electromotive force and the transient field parameters. By analyzing the error conditions trained by different numbers of neurons in a single hidden layer, the hidden layers of the 5-layer deep neural network are determined. The number of layered neurons is 13,8,5,8,13. The training algorithm chooses the improved Nadam algorithm with adaptive learning rate, which can speed up the training process. The trained deep neural network model is simulated and verified by a typical electrical model, and it is found that it has a good response to different geoelectric models, which proves the feasibility of calculating apparent resistivity based on deep learning put forward in this paper. The actual application results show that the trained deep neural network model can quickly and accurately calculate the apparent resistivity, and its effectiveness is verified by drilling.

Key words： TEM deep learning artificial neural network whole-region apparent resistivity

收稿日期: 2020-11-08 出版日期: 2021-07-27

P631

基金资助:新疆维吾尔自治区自然科学基金项目(2017D01C064);新疆维吾尔自治区研究生科研创新项目(XJ2020G044)

通讯作者: 张莹莹

作者简介: 吴国培(1994-),男,在读硕士,研究方向为瞬变电磁探测。Email: wu_guopei@163.com

引用本文:

吴国培, 张莹莹, 张博文, 赵华亮. 基于深度学习的中心回线瞬变电磁全区视电阻率计算[J]. 物探与化探, 2021, 45(3): 750-757.
WU Guo-Pei, ZHANG Ying-Ying, ZHANG Bo-Wen, ZHAO Hua-Liang. The calculation of full-region apparent resistivity of central loop TEM based on deep learning. Geophysical and Geochemical Exploration, 2021, 45(3): 750-757.

链接本文:

https://www.wutanyuhuatan.com/CN/10.11720/wtyht.2021.1511 或 https://www.wutanyuhuatan.com/CN/Y2021/V45/I3/750

Fig.1 F(u)与瞬变场参数u的变化特征

Fig.2 DNN计算视电阻率选取判别

Fig.3 通用5层DNN结构^[23]

Table 1 人工神经网络隐含层不同神经元个数的误差情况

Fig.4 深度神经网络Nadam训练算法误差下降

Fig.5 不同训练误差DNN视电阻率对比

Fig.6 4种地电模型视电阻率与视深度示意

Fig.7 和静水资源勘察视电阻率拟断面

[1]	朴化荣. 电磁测深法原理[M]. 北京: 地质出版社, 1990.
[1]	Piao H R. The theory of electromagnetic sounding[M]. Beijing: Geology Publishing House, 1990.
[2]	方文藻, 李予国, 李貅. 瞬变电磁测深法原理[M]. 西安: 西北工业大学出版社, 1993.
[2]	Fang W Z, Li Y G, Li X. Principle of transient electromagnetic sounding[M]. Xi’an: Northwestern Polytechnical University Press, 1993.
[3]	牛之琏. 时间域电磁法原理[M]. 长沙: 中南工业大学出版社, 1992.
[3]	Niu Z L. The theory of time-domain electromagnetic methods[M]. Changsha: Central South University of Technology Press, 1992.
[4]	蒋邦远. 实用近区磁源瞬变电磁法勘探[M]. 北京: 地质出版社, 1998.
[4]	Jiang B Y. Practical near zone source transient em sounding[M]. Beijing: Geological Publishing House, 1998.
[5]	李貅. 瞬变电磁测深的理论与应用[M]. 西安: 陕西科学技术出版社, 2002:1195-1200.
[5]	Li X. The theory and application of transient electromagnetic sounding[M]. Xi’an: Shanxi Science Technology Press, 2002:1195-1200.
[6]	苏朱刘, 胡文宝. 中心回线方式瞬变电磁测深虚拟全区视电阻率和一维反演方法[J]. 石油物探, 2002, 41(2):216-221.
[6]	Su Z L, Hu W B. Pseudo full-region apparent resistivity and its one-dim ensional inversion for center loop line configuration TEM data[J]. Geophysical Prospecting for Petroleum, 2002, 41(2):216-221.
[7]	白登海, Meju Maxwell-A, 卢健, 等. 时间域瞬变电磁法中心方式全程视电阻率的数值计算[J]. 地球物理学报, 2003, 46(5):697-704.
[7]	Bai D H, Meju M A, Lu J, et al. Numerical calculation of all-time apparent resistivity for the central loop transient electromagnetic method[J]. Chinese Journal of Geophys., 2003, 46(5):697-704.
[8]	熊彬. 大回线瞬变电磁法全区视电阻率的逆样条插值计算[J]. 吉林大学学报:地球科学版, 2005, 35(4):515-519.
[8]	Xiong B. Inverse spline interpolation for the calculation of all-time resistivity for the large loop transient electromagnetic method[J]. Journal of Jilin University:Earth Science Edition, 2003, 35(4):515-519.
[9]	陈清礼. 瞬变电磁法全区视电阻率的二分搜索算法[J]. 石油天然气学报, 2009, 31(2):45-49.
[9]	Chen Q L. Searching algorithm for full time apparent resistivity from TEM electromotive force data[J]. Journal of Oil and Gas Technology, 2009, 31(2):45-49.
[10]	武军杰, 王兴春, 邓晓红, 等. 定源回线瞬变电磁x分量视电阻率计算方法[J]. 物探与化探, 2012, 36(4):684-687.
[10]	Wu J J, Wang X C, Deng X H, et al. The method for Calculating x-component apparent resistivity of TEM fixed loop [J]. Geophysical and Geochemical Exploration, 2012, 36(4):684-687.
[11]	郝延松, 胡博, 于润桥, 等. 磁性源瞬变电磁法视电阻率计算方法[J]. 物探与化探, 2012, 36(6):1034-1039.
[11]	Hao Y S, Hu B, Yu R Q, et al. The calculation of apparent resistivity for magnetic source TEM[J]. Geophysical and Geochemical Exploration, 2012, 36(6):1034-1039.
[12]	薛国强, 李貅, 底青云. 瞬变电磁法正反演问题研究进展[J]. 地球物理学进展, 2008, 90(4):1165-1172.
[12]	Xue G Q, Li X, Di Q Y. Research progress in TEM forward modeling and inversion calculation[J]. Progress in Geophysics, 2008, 90(4):1165-1172.
[13]	谢林涛, 付志红, 谢品芳, 等. 基于神经网络的视电阻率快速算法[J]. 地球物理学进展, 2009, 24(4):1527-1532.
[13]	Xie L T, Fu Z H, Xie P F, et al. A fast a1gorithm of apparent resistivity based on neural networks[J]. Progress in Geophysics, 2009, 24(4):1527-1532.
[14]	谢品芳, 谢林涛, 付志红, 等. 基于BP网络的视电阻率快速计算[J]. 重庆大学学报, 2009, 32(9):1111-1116.
[14]	Xie P F, Xie L T, Fu Z H, et al. Fast algorithm for apparent resistivity calculation based on BP neural networks[J]. Journal of Chongqing University, 2009, 32(9):1111-1116.
[15]	Hinton G, Salakhutdinov R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786):504-507. pmid: 16873662
[16]	余凯, 贾磊, 陈雨强, 等. 深度学习的昨天、今天和明天[J]. 计算机研究与发展, 2013, 50(9):1799-1804.
[16]	Yu K, Jia L, Chen Y Q, et al. Yesterday, today and tomorrow of deep learning[J]. Journal of Computer Research and Development, 2013, 50(9):1799-1804.
[17]	毛健, 赵红东, 姚婧婧. 人工神经网络的发展及应用[J]. 电子设计工程, 2011, 19(24):62-65.
[17]	Mao J, Zhao H D, Yao J J. Development and application of artificial neural networks[J]. Electronic Design Engineering, 2011, 19(24):62-65.
[18]	Dozat T. Incorporating nesterov momentum into adam[J]. ICLR Workshop, 2016, 20(1):2013-2016.
[19]	秦善强, 付志红, 朱学贵, 等. 遗传神经网络的瞬变电磁视电阻率求解算法[J]. 电工技术学报, 2017, 32(12):146-154, 250.
[19]	Qin S Q, Fu Z H, Zhu X G, et al. Genetic neural network for apparent resistivity solution of transient electromagnetic[J]. Transactions of China Electrotechnical Society, 2017, 32(12):146-154,250.
[20]	薛国强. 论瞬变电磁测深法的探测深度[J]. 石油地球物理勘探, 2004, 39(5):575-578.
[20]	Xue G Q. On the detection depth of the transient electromagnetic sounding method[J]. Petroleum Geophysical Prospecting, 2004, 39(5):575-578.
[21]	吕国印. 瞬变电磁法的现状与发展趋势[J]. 物探化探计算技术, 2007, 29(S1):111-115.
[21]	Lyu G Y. The status quo and development trend of transient electromagnetic method[J]. Computing Techniques for Geophysical and Geochemical Exploration, 2007, 29(S1):111-115.
[22]	刘天佑. 地球物理勘探概论[M]. 北京: 地质出版社, 2007.
[22]	Liu T Y. Introduction to geophysical prospecting[M]. Beijing: Geological Publishing House, 2007.
[23]	Nabighian M N. Quasi-static transient response of a conducting half-space-an approximate representation[J]. Geophysics, 1979, 44(10):1700-1705. doi: 10.1190/1.1440931
[24]	Francois Chollet. Python深度学习[M]. 张亮, 译. 北京: 人民邮电出版社, 2018.
[24]	Francois chollet. Python deep learning[M]. Zhang L, translated. Beijing: People’s Posts and Telecommunications Press, 2018.
[25]	毛勇华, 桂小林, 李前, 等. 深度学习应用技术研究[J]. 计算机应用研究, 2016, 33(11):3201-3205.
[25]	Mao Y H, Gui X L, Li Q, et al. Research on deep learning application technology[J]. Application Research of Computers, 2016, 33(11):3201-3205.
[26]	牟晋娟. 深度神经网络中激活函数的研究[J]. 电脑编程技巧与维护, 2019, 414(12):59-61.
[26]	Mou J J. Research on activation function in deep neural networks[J]. Computer Programming Skills & Maintenance, 2019, 414(12):59-61.
[27]	张崇荣. 基于卷积神经网络的图像分类方法研究[D]. 成都:电子科技大学, 2019.
[27]	Zhang C R. Research on image classification method based on convolutional neural network[D]. Chengdu: University of Electronic Science and Technology of China, 2019.
[28]	吴昌友. 神经网络的研究及应用[D]. 哈尔滨:东北农业大学, 2007.
[28]	Wu C Y. Research and application of neural network[D]. Harbin: Northeast Agricultural University, 2007.
[29]	柳小桐. BP神经网络输入层数据归一化研究[J]. 机械工程与自动化, 2010, 160(3):122-123, 126.
[29]	Liu X T. Study on data normalization in BP neural network[J]. Mechanical Engineering & Automation, 2010, 160(3):122-123.
[30]	沈花玉, 王兆霞, 高成耀, 等. BP神经网络隐含层单元数的确定[J]. 天津理工大学学报, 2008, 91(5):13-15.
[30]	Shen H Y, Wang Z X, Gao C Y, et al. Determining the number of BP neural network hidden layer units[J]. Journal of Tianjin University of Technology, 2008, 91(5):13-15.

[1]	陈学群, 李成光, 田婵娟, 刘丹, 辛光明, 管清花. 高密度电阻率法在咸水入侵监测中的应用[J]. 物探与化探, 2021, 45(5): 1347-1353.
[2]	朱颜, 韩向义, 岳欣欣, 杨春峰, 常文鑫, 邢丽娟, 廖晶. 致密砂岩储层脆性测井评价方法研究及应用——以鄂尔多斯盆地渭北油田为例[J]. 物探与化探, 2021, 45(5): 1239-1247.
[3]	邢涛, 袁伟, 李建慧. 回线源瞬变电磁法的一维Occam反演[J]. 物探与化探, 2021, 45(5): 1320-1328.
[4]	李瑞友, 张淮清, 吴昭. 基于在线惯序极限学习机的瞬变电磁非线性反演[J]. 物探与化探, 2021, 45(4): 1048-1054.
[5]	裴肖明, 冯国瑞, 戚庭野. 瞬变电磁法探测复杂状态下煤矿充水采空区物理模拟实验[J]. 物探与化探, 2021, 45(4): 1055-1063.
[6]	刘伟, 黄韬, 王庭勇, 刘怡, 张继, 刘文涛, 张琦斌, 李强. 综合物探方法在城市隐伏断裂探测中的应用[J]. 物探与化探, 2021, 45(4): 1077-1087.
[7]	张莹莹. 电性源瞬变电磁法综述[J]. 物探与化探, 2021, 45(4): 809-823.
[8]	王文杰, 郝一, 薄海军, 王海龙, 徐浩清, 李永利, 毛磊, 刘永新, 袁帅. 包头市固阳县矿集区高密度电阻率法找水定井实例分析[J]. 物探与化探, 2021, 45(4): 869-881.
[9]	张勇, 马晓东, 李彦婧, 蔡景顺. 深度学习在南川页岩气含气量预测中的应用[J]. 物探与化探, 2021, 45(3): 569-575.
[10]	陈小龙, 高坡, 程顺达, 王晓青, 罗可. 西藏帮浦东段—笛给铅锌矿区CSAMT异常特征与深部找矿预测[J]. 物探与化探, 2021, 45(2): 361-368.
[11]	陈健强, 李雁川, 田浩, 李汉超. 含水采空区全空间瞬变电磁响应分析[J]. 物探与化探, 2021, 45(2): 546-550.
[12]	张莹莹. 多辐射场源半航空瞬变电磁法多分量响应特征分析[J]. 物探与化探, 2021, 45(1): 102-113.
[13]	李展辉, 杨淼鑫, 曹学峰. 瞬变电磁法感应电压场与B场探测效果的数值计算对比分析[J]. 物探与化探, 2021, 45(1): 114-126.
[14]	梁立锋, 刘秀娟, 张宏兵, 陈程浩, 陈锦华. 超参数对GRU-CNN混合深度学习弹性阻抗反演影响研究[J]. 物探与化探, 2021, 45(1): 133-139.
[15]	罗维斌, 丁志军, 高曙德, 张星. *测量磁场水平分量H_y的电性源广域电磁测深法*[J]. 物探与化探, 2021, 45(1): 46-56.

Viewed

Full text

Abstract

Cited

Shared

Discussed