Please wait a minute...
E-mail Alert Rss
 
物探与化探  2021, Vol. 45 Issue (4): 1048-1054    DOI: 10.11720/wtyht.2021.1514
  方法研究·信息处理·仪器研制 本期目录 | 过刊浏览 | 高级检索 |
基于在线惯序极限学习机的瞬变电磁非线性反演
李瑞友(), 张淮清(), 吴昭
重庆大学 输配电装备及系统安全与新技术国家重点实验室,重庆 400044
Online sequential extreme learning machine for transient electromagnetic nonlinear inversion
LI Rui-You(), ZHANG Huai-Qing(), WU Zhao
State Key Laboratory of Power Transmission Equipment & System Security and New Technology, Chongqing University, Chongqing 400044, China
全文: PDF(2487 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 

基于梯度下降法的传统人工神经网络瞬变电磁反演方法计算效率低,不能保证全局收敛。为了解决上述问题,提出一种在线惯序极限学习机(online sequential extreme learning machine, OSELM)的瞬变电磁反演方法。该方法针对瞬变电磁法所获取的高维勘探数据进行建模反演,首先,通过随机设定隐层参数(输入权值和偏差)来简化模型的学习过程;然后,将测试得到的预测样本加入训练样本中,作为下一次的更新信息,建立在线贯序极限学习机预测模型,从而最大限度提高反演精度;最后,设计了两个经典的瞬变电磁层状地电模型并进行了拟二维地电模型的反演。反演结果表明,该方法能够较好地解决瞬变电磁法高维数据非线性建模的反演问题,同时相较极限学习机(extreme learning machine, ELM),非线性反演方法具有更加准确的反演结果、更好的泛化能力以及更高的计算效率,为神经网络在地球物理反演中的应用提供了新思路。

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
李瑞友
张淮清
吴昭
关键词 瞬变电磁法人工神经网络在线惯序极限学习机反演    
Abstract

The traditional transient electromagnetic inversion method using artificial neural network based on gradient descent method is inefficient and can not guarantee global convergence. In order to solve these problems, this paper proposes a transient electromagnetic inversion method based on on online sequential extreme learning machine (OSELM). This approaches is used for inversion of high-dimensional exploration data obtained by transient electromagnetic method. Firstly, the hidden layer parameters (input weight and deviation) are randomly set to simplify the learning process of the model. Then, the prediction samples obtained from the test are added to the training samples as the next update information, and the online sequential extreme learning machine prediction model is established to maximize the inverse accuracy. Finally, the inversion results of two classical TEM layered geoelectric models and a quasi two-dimensional geoelectric model show that the proposed method can solve the problem of nonlinear modeling and high-dimensional data for TEM inversion, and a comparison with extreme learning machine (ELM) shows that this method has more accurate inversion, better generalization ability and higher calculation efficiency, which provides a new idea for the application of neural network in geophysical inversion.

Key wordstransient electromagnetic method    artificial neural network    online sequential extreme learning machine    inversion
收稿日期: 2020-11-23      出版日期: 2021-08-20
:  P631  
基金资助:国家自然科学基金(51377174)
通讯作者: 张淮清
作者简介: 李瑞友(1994-),男, 博士,毕业于哈尔滨理工大学,主要从事瞬变电磁正反演理论及人工智能研究工作。 Email: 1378546842@qq.com
引用本文:   
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.
LI Rui-You, ZHANG Huai-Qing, WU Zhao. Online sequential extreme learning machine for transient electromagnetic nonlinear inversion. Geophysical and Geochemical Exploration, 2021, 45(4): 1048-1054.
链接本文:  
https://www.wutanyuhuatan.com/CN/10.11720/wtyht.2021.1514      或      https://www.wutanyuhuatan.com/CN/Y2021/V45/I4/1048
Fig.1  层状地电模型和TEM方法示意
五层模型 ρ/(Ω·m) h/m
ρ1 ρ2 ρ3 ρ4 ρ5 h1 h2 h3 h4
pmin 10 5 1 20 100 10 10 5 10
pmax 1 000 500 100 2000 10 000 1 000 1000 500 10 000
九层模型 ρ/(Ω·m) h/m
ρ1 ρ2 ρ3 ρ4 ρ5 ρ6 ρ7 ρ8 ρ9 h1 h2 h3 h4 h5 h6 h7 h8
pmin 10 1 10 1 10 5 13 10 10 2 2 2 2 6 2 3 1
pmax 1 000 100 1 000 100 1 000 500 1 300 1 000 1 000 200 200 200 200 600 200 300 100
Table 1  5层、9层地电模型各层参数的最大值与最小值
Fig.2  ELM神经网络结构
Fig.3  OSELM算法计算流程
Algorithm 5层模型 9层模型
R2 RRMSE APE/% Time/s R2 RRMSE APE/% Time/s
OSELM 0.9978 0.1148 6.465 0.0089 0.9975 0.1542 8.372 0.0089
ELM 0.9978 0.1540 8.174 0.0097 0.9974 0.1904 10.218 0.0099
Table 2  两种极限学习机方法反演性能的比较
五层模型 ρ/(Ω·m) h/m
ρ1 ρ2 ρ3 ρ4 ρ5 h1 h2 h3 h4
理论值 100 50 10 200 1000 100 100 50 100
九层模型 ρ/(Ω·m) h/m
ρ1 ρ2 ρ3 ρ4 ρ5 ρ6 ρ7 ρ8 ρ9 h1 h2 h3 h4 h5 h6 h7 h8
理论值 100 10 100 10 100 50 130 100 100 20 20 20 20 60 20 30 10
Table 3  5层、9层地电模型反演理论值
Fig.4  5层地电模型(a)与9层地电模型(b)的不同算法反演结果
Fig.5  5层地电模型(a)与9层地电模型(b)的正演响应曲线
Fig.6  拟二维模型示意图及测量位置
Fig.7  拟二维地电模型OSELM方法(a)和ELM方法(b)的反演结果
[1] He Z X, Zhao Z, Liu H Y, et al. TFEM for oil detection: Case studies[J]. Leading Edge, 2012,31(5):518-521. doi: 10.1190/tle31050518.1.
[2] Tantum S L, Collins L M. A comparison of algorithms for subsurface target detection and identification using time-domain electromagnetic induction data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2001,39(6):1299-1306. doi: 10.1109/36.927453.
[3] 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, 2008,23(4):201-208.
[3] Xue G Q, Li X, Di Q Y. Research progress on forward and inversion of transient electromagnetic method[J]. Progress in Geophysics, 2008,23(4):201-208.
[4] Loke M H, Barker R D. Least-squares deconvolution of apparent resistivity pseudosections[J]. Geophysics, 1995,60(6):1682-1690. doi: 10.1190/1.1443900.
[5] 张继令, 翁爱华. 中心回线瞬变电磁测深一维Occam反演[J]. 铁道勘察, 2007,33(2):40-43.
[5] Zhang J L, Weng A H. One dimensional Occam inversion of transient electromagnetic sounding in central loop[J]. Railway Survey, 2007,33(2):40-43.
[6] 杨海燕, 李锋平, 岳建华, 等. 基于“烟圈”理论的圆锥型场源瞬变电磁优化反演[J]. 中国矿业大学学报, 2016,45(6):1230-1237.
[6] Yang H Y, Li F P, Yue J H, et al. Transient electromagnetic optimization inversion of conical field source based on “smoke circle” theory[J]. Journal of China University of Mining and Technology, 2016,45(6):1230-1237.
[7] Puzyrev V, Swidinsky A. Inversion of 1D frequency-and time-domain electromagnetic data with convolutional neural networks[J]. Computers and Geosciences, 2020,149:104681. doi: 10.1016/j.cageo.2020.104681.
[8] 贺懿, 张进, 刘怀山. 基于神经网络的面波迭代反演应用研究[J]. 西南石油大学学报:自然科学版, 2010,32(1):40-44.
[8] He Y, Zhang J, Liu H S. Application of surface wave iterative inversion based on neural network[J]. Journal of Southwest Petroleum University:Natural Science Edition, 2010,32(1):40-44.
[9] 徐海浪, 吴小平. 电阻率二维神经网络反演[J]. 地球物理学报, 2006,49(2) : 584-589.
[9] Xu H L, Wu X P. Two dimensional neural network inversion of resistivity[J]. Chinese Journal of Geophysics, 2006,49(2) : 584-589.
[10] 嵇艳鞠, 徐江, 吴琼, 等. 基于神经网络电性源半航空视电阻率反演研究[J]. 电波科学学报, 2014,29(5):973-980.
[10] Ji Y J, Xu J, Wu Q, et al. Semi aerial apparent resistivity inversion based on neural network electrical source[J]. Journal of Radio Science, 2014,29(5):973-980.
[11] Srinivas Y, Raj A S, Oliver D H, et al. A robust behavior of Feed Forward Back propagation algorithm of Artificial Neural Networks in the application of vertical electrical sounding data inversion[J]. Geoscience Frontiers, 2012,3(5):729-736.
[12] Maiti S, Erram V C, Gupta G, et al. ANN based inversion of DC resistivity data for groundwater exploration in hard rock terrain of western Maharashtra (India)[J]. Journal of Hydrology, 2012,464:294-308. doi: 10.1016/j.jhydrol.2012.07.020.
[13] Johnson O L, Aizebeokhai A P. Application of artificial neural network for the inversion of electrical resistivity data[J]. Journal of Informatics and Mathematical Sciences, 2017,9(2):297-316.
[14] Wilamowski B M, Yu H. Neural network learning without backpropagation[J]. IEEE Transactions on Neural Networks, 2010,21(11):1793-1803. doi: 10.1109/TNN.2010.2073482.
[15] Huang G B, Zhu Q Y, Siew C K. Extreme learning machine: a new learning scheme of feedforward neural networks[C]// IEEE International Joint Conference on Neural Networks. IEEE, 2005.
[16] Cai Z, Gu J, Luo J, et al. Evolving an optimal kernel extreme learning machine by using an enhanced grey wolf optimization strategy[J]. Expert Systems with Applications, 2019,138. doi: 10.1016/j.eswa.2019.07.031.
[17] 王保义, 赵硕, 张少敏. 基于云计算和极限学习机的分布式电力负荷预测算法[J]. 电网技术, 2014,38(2):526-531. doi: 10.13335/j.1000-3673.pst.2014.02.039.
[17] Wang B Y, Zhao S, Zhang S M. Distributed power load forecasting algorithm based on cloud computing and extreme learning machine[J]. Power System Technology, 2014,38(2):526-531. doi: 10.13335/j.1000-3673.pst.2014.02.039.
[18] She Q, Chen K, Ma Y, et al. Sparse representation-based extreme learning machine for motor imagery EEG classification[J]. Computational Intelligence & Neuroscience, 2018(1):1-9. doi: 10.1190/tle31050518.1.
[19] 魏迪, 刘德山, 闫德勤, 等. 应用于人脸图像识别的邻域保持极限学习机[J]. 计算机工程与应用, 2019,55(11):187-191.
[19] Wei D, Liu D S, Yan D Q, et al. Neighborhood preserving extreme learning machine for face image recognition[J]. Computer Engineering and Applications, 2019,55(11):187-191.
[20] Li H F, Wang S H, Zou Z, et al. An integrated methodology for rule extraction from ELM—Based vacuum tank degasser multiclassifier for decision-making[J]. Energies, 2019,12(18):3535. doi: 10.3390/en12183535.
[21] Liang N Y, Huang G B, Saratchandran P, et al. A fast and accurate online sequential learning algorithm for feedforward networks[J]. IEEE Transactions on Neural Networks, 2006,17:1411-1423. doi: 10.1109/TNN.2006.880583.
[22] 裴飞. 基于在线序列极限学习机的变压器故障诊断研究[D]. 北京:华北电力大学, 2015.
[22] Pei F. Research on transformer fault diagnosis based on online sequential extreme learning machine[D]. Beijing: North China Electric Power University, 2015.
[23] 张明洋, 闻英友, 杨晓陶, 等. 一种基于增量加权平均的在线序贯极限学习机算法[J]. 控制与决策, 2017,32(10):1887-1893.
[23] Zhang M Y, Wen Y Y, Yang X T, et al. An online sequential extreme learning machine algorithm based on incremental weighted average[J]. Control and Decision Making, 2017,32(10):1887-1893.
[24] Yin J C, Zou Z J, Xu F, et al. Online ship roll motion prediction based on grey sequential extreme learning machine[J]. Neurocomputing, 2014,129(10):168-174.
[25] Kaufman A A, Keller G V. Frequency and transient sounding[J]. Elsevier Methods in Geochemistry & Geophysics, 1983,21. doi: 10.1016/0031-9201(83)90051-1.
[26] Vignoli G, Fiandaca G, Christiansen A V, et al. Sharp spatially constrained inversion with applications to transient electromagnetic data[J]. Geophysical Prospecting, 2014,63(1):243-255. doi: 10.1111/1365-2478.12185.
[1] 冯旭亮, 魏泽坤. 基于界面反演增强的位场边缘识别方法[J]. 物探与化探, 2022, 46(1): 130-140.
[2] 黄远生, 王彦国, 罗潇. 基于归一化磁源强度垂向差分的磁源参数快速估计方法[J]. 物探与化探, 2021, 45(6): 1588-1596.
[3] 何可, 郭明, 胡章荣, 易国财, 王仕兴. 半航空瞬变电磁L1范数自适应正则化反演[J]. 物探与化探, 2021, 45(5): 1338-1346.
[4] 郭培虹, 冯治汉, 王万银, 唐小平, 刘生荣. 北秦岭华阳川地区重磁三维反演及岩浆岩特征研究[J]. 物探与化探, 2021, 45(5): 1217-1225.
[5] 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.
[6] 刘鸿洲, 王孟华, 张浩, 彭玲丽, 李雯, 张杰, 赵智鹏, 伍泽荆. 基于分频构形反演方法的河道砂精准预测——以华北冀中探区赵皇庄地区为例[J]. 物探与化探, 2021, 45(5): 1311-1319.
[7] 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.
[8] 张鹏飞, 张世晖. 西湖凹陷平湖组砂泥岩岩性神经网络地震预测[J]. 物探与化探, 2021, 45(4): 1014-1020.
[9] 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.
[10] 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.
[11] 张莹莹. 电性源瞬变电磁法综述[J]. 物探与化探, 2021, 45(4): 809-823.
[12] 刘辉, 李静, 曾昭发, 王天琪. 基于贝叶斯理论面波频散曲线随机反演[J]. 物探与化探, 2021, 45(4): 951-960.
[13] 马良涛, 范廷恩, 许学良, 董建华, 蔡文涛. 油气检测多技术联合在B油田的应用研究[J]. 物探与化探, 2021, 45(4): 961-969.
[14] 李英宾. 可控源音频大地电磁测量对腾格尔坳陷东北缘下白垩统赛汉组砂体的识别及其地质意义[J]. 物探与化探, 2021, 45(3): 616-623.
[15] 李凡, 周明, 李开天, 鲁恺, 李振宇. 单斜地形情况下的磁共振测深方法反演研究[J]. 物探与化探, 2021, 45(3): 712-725.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
京ICP备05055290号-3
版权所有 © 2021《物探与化探》编辑部
通讯地址:北京市学院路29号航遥中心 邮编:100083
电话:010-62060192;62060193 E-mail:whtbjb@sina.com