Please wait a minute...
E-mail Alert Rss
 
物探与化探  2020, Vol. 44 Issue (2): 394-400    DOI: 10.11720/wtyht.2020.1504
  方法研究·信息处理·仪器研制 本期目录 | 过刊浏览 | 高级检索 |
基于卷积神经网络识别重力异常体
王逸宸1,2,3, 柳林涛1,2, 许厚泽1,2
1. 中国科学院 测量与地球物理研究所,湖北 武汉 430077
2. 中国科学院 大地测量与地球动力学国家重点实验室,湖北 武汉 430077
3. 中国科学院大学 地球与行星科学学院,北京 100049
The identification of gravity anomaly body based on the convolutional neural network
Yi-Chen WANG1,2,3, Lin-Tao LIU1,2, Hou-Ze XU1,2
1. Institute of Geodesy and Geophysics, Chinese Academy of Sciences, Wuhan 430077, China
2. State Key Laboratory of Geodesy and Earth’s Dynamics, Chinese Academy of Sciences, Wuhan 430077, China
3. Collage of Earth and Planetary Sciences,University of Chinese Academy of Sciences, Beijing 100049, China
全文: PDF(1573 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 

本文将深度学习与重力异常体识别结合,基于近年来在图像识别邻域取得优异效果的卷积神经网络,将重力观测等值线图看作待识别的二维图像,将地下重力异常体的空间参数看作识别输出,从而形成适用于异常体识别的卷积神经网络模型。在训练中,随机生成大量不同参数的三维异常体模型,正演得到其重力观测二维数据,用异常体模型参数标签和重力数据训练卷积神经网络。在模型算例中测试训练好的网络模型,其识别准确性良好。同时,相比于传统神经网络从二维重力测线中识别异常体的埋深,卷积神经网络可从二维的重力数据识别三维异常体的埋深和大小信息。最后,将网络应用于澳大利亚Kauring地区重力观测数据,异常体识别结果与前人研究结果相符。说明卷积神经网络具泛化能力,可用于识别实测重力异常体,结果可靠。

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
王逸宸
柳林涛
许厚泽
关键词 深度学习卷积神经网络重力异常体识别参数反演    
Abstract

This study combines the deep learning with the identification of gravity anomaly body. Based on the CNN (convolutional neural network) which has been gaining its use in the past several years in the field of image identification, the contour image of gravity signal is taken as the unidentified image, while the space parameters of the gravity anomaly body will be identified through CNN. In the training phase, a large number of the 3D anomaly bodies are generated with random variation of parameters, then the network is fed with parametric labels and the computed gravity contour images. The testing is performed with generated testing models to estimate the performance of the trained model. The trained CNN accuracy shows excellent accuracy in the identifications. Then the CNN model is tested with measured main gravity anomaly data of Kauring area in West Australia, and the identified parameters of the 3D anomaly body are compared with known results. It is shown that the generalization of CNN can handle identification of the measured gravity data.

Key wordsdeep learning    convolutional neural network    identification of gravity anomaly body    parametric inversion
收稿日期: 2019-10-28      出版日期: 2020-04-22
:  P631  
基金资助:国家自然科学基金项目(Y211641064);国家重大科学仪器设备开发专项基金项目“海洋/航空重力仪研制”(20011YQ120045)
作者简介: 王逸宸(1988-),男,博士生,主要研究方向为重力反演算法。 Email: ycwang_1988@asch.whigg.ac.cn
引用本文:   
王逸宸, 柳林涛, 许厚泽. 基于卷积神经网络识别重力异常体[J]. 物探与化探, 2020, 44(2): 394-400.
Yi-Chen WANG, Lin-Tao LIU, Hou-Ze XU. The identification of gravity anomaly body based on the convolutional neural network. Geophysical and Geochemical Exploration, 2020, 44(2): 394-400.
链接本文:  
https://www.wutanyuhuatan.com/CN/10.11720/wtyht.2020.1504      或      https://www.wutanyuhuatan.com/CN/Y2020/V44/I2/394
Fig.1  卷积神经网络结构
Fig.2  用于识别重力异常体的卷积神经网络结构
Fig.3  卷积神经网络训练损失函数
a—识别深度;b—识别线度
Fig.4  用于测试神经网络的异常体
Fig.5  测试异常体的重力
a—模型a;b—模型b;c—模型c; d—模型d
模型编号 真实深度、线度/m 识别深度、线度/m
a 300.0, 200.0 307.7, 201.5
b 600.0, 200.0 621.6, 201.7
c 400.0, 252.0 398.0, 249.1
d 400.0, 252.0 392.7, 251.0
Table 1  测试异常体识别结果
Fig.6  Kauring试验场的重力异常
作者 方法类别 异常体深度、线度/m
Martinez et al(2012) 等效源识别 250.0, 300.0
Liu et al.(2015) 密度反演 420.0, 120.0
田宇(2019) 密度反演 400, 160
本文 深度学习识别 421.6, 148.8
Table 2  Kauring测区异常体识别结果与前人对比
[1] Reading A M, Cracknell M J, Bombardieri D J , et al. Combining machine learning and geophysical inversion for applied geophysics[J]. ASEG Extended Abstracts, 2019,2015(1):1-5.
[2] Aster R C, Borchers B, Thurber C H . Parameter estimation and inverse problems[M]. Amsterdam, Netherlands: Elsevier, 2005.
[3] Spichak V, Popova I . Methodology of neural network inversion of geophysical data[J]. Izvestiya, Physics of the Solid Earth, 2005,41(3):241-254.
[4] Grauch V J S . Detection of cavities and tunnels from gravity data using a neural network[J]. Exploration Geophysics, 2018,32(3-4):204-208.
[5] 周洪祥 . 基于神经网络的重力异常反演[D]. 成都:成都理工大学, 2014.
[5] Zhou H X . Gravity anomaly inversion based on neural networks[D]. Chengdu: Chengdu University of Technology, 2014.
[6] 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.
[6] Guan Z N, Hou J S, Huang L P , et al. Inversion of gravity and magnetic anomalies using pseduo-bp neural network method and its application[J]. Journal of Geophysics, 1998,41(2):242-251.
[7] 相鹏 . 基于拟径向基函数神经网络的重力反演方法[C]//2018年中国地球科学联合学术年会, 2018: 933-935.
[7] Xiang P . Gravity inversion method based on quasi radial basis function neural network[C]//2018 China Geoscience Union Annual Meeting, 2018: 933-935.
[8] 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.
[8] Geng M X, Yang Q J . 2-D density inversion with the RBF neural network method[J]. Oil Geophysical Prospecting. 2013,48(4):651-657.
[9] Al-Garni M A . Inversion of residual gravity anomalies using neural network[J]. Arabian Journal of Geosciences, 2013,6(5):1509-1516.
[10] Al-Garni M A . Interpretation of spontaneous potential anomalies from some simple geometrically shaped bodies using neural network inversion[J]. Acta Geophysica, 2010,58(1):143-162.
[11] El-Kaliouby H, Al-Garni M . Inversion of self-potential anomalies caused by 2D inclined sheets using neural networks[J]. Journal of Geophysics and Engineering, 2009,6(1):29-34.
[12] Ucan O N, Osman O, Albora A M . A new approach for residual gravity anomaly profile interpretations: Forced Neural Network (FNN)[J]. Annals of Geophysics, 2006,49(6):1201-1208.
[13] 任强强, 王跃钢, 腾红磊 , 等. 基于小波神经网络的盲区重力数据预测[J]. 大地测量与地球动力学, 2016,36(4):359-363.
[13] Ren Q Q, Wang Y G, Teng H L , et al. The gravity data forecast of unmeasurable zone based on wavelet neural network[J]. Journal of Geodesy and Geodynamics, 2016,36(4):359-363.
[14] Bengio Y, Courville A, Vincent P . Representation learning: a review and new perspectives[D]. Montreal Canada: University of Montreal, 2012.
[15] 周志华 . 机器学习[M]. 北京: 清华大学出版社, 2016.
[15] Zhou Z H. Machine learning[M]. Beijing: Tsinghua University Press, 2016.
[16] Goodfellow I, Bengio Y, Courville A. Deep Learning[M]. Boston: MIT Press, 2016.
[17] Dan Cireş D, Meier U, Schmidhuber J. Multi-column deep neural networks for image classification [C]// Providence, USA: 2012 IEEE Conference on Computer Vision and Pattern Recognition, 2012: 3642-3649.
[18] Cun L, Boser Y, Denker B , et al. Back-propagation applied to Handwritten Zip-code recognition[J]. Neural Computation-NECO, 1992,1(4), 541-551.
[19] Fukushima K . Neocognitron[J]. Scholarpedia, 2007: 1(2):1717.
[20] Krizhevsky A, Sutskever I, Hinton G. ImageNet classification with deep convolutional neural networks [C]// New York: ACM, 2012: 84-90.
[21] Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions [C]//Boston: 2015 IEEE Conference on Computer Vision and Pattern Recognition, 2015: 1-9.
[22] Russakovsky O, Deng J, Su H , et al. ImageNet large scale visual recognition challenge[J]. International Journal of Computer Vision, 2014,115(3), 211-252.
[23] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition [C]//Boston: 2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
[24] Ciresan D, Meier U, Masci J, et al. Flexible, high performance convolutional neural networks for image classification [C]// Barcelona: Proceedings of the 22nd International Joint Conference on Artificial Intelligence, 2011: 1237-1242.
[25] Levi G, Hassnet T. Age and gender classification using convolutional neural networks [C]//Boston: 2015 IEEE Conference on Computer Vision and Pattern Recognition, 2015: 34-42.
[26] Laina I, Rupprecht C, Belagiannis V, et al. Deeper depth prediction with fully convolutional residual networks [C]//Boston: 2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016: 239-248.
[27] Martinez C, Li Y. Understanding gravity gradiometry processing and interpretation through the Kauring test site data [C]// Brisbane: 22nd International Geophysical Conference and Exhibition, 2012: 26-29.
[28] Liu J Z, Liu L T, Liang X H , et al. 3D density inversion of gravity gradient data using the extrapolated Tikhonov regularization[J]. Applied Geophysics, 2015,12(2):137-146.
[29] 田宇, 柯小平, 王勇 . 利用航空重力梯度反演Kauring试验场三维密度结构[J]. 武汉大学学报:信息科学版, 2019,44(4):501-509.
[29] Tian Y, Ke X P, Wang Y . Inversion of three-dimensional density structure using airborne gradiometry data in Kauring test site[J]. Geomatics and Information Science of Wuhan University, 2019,44(4):501-509.
[1] 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.
[2] 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.
[3] 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.
[4] 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.
[5] 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.
[6] 王亚, 易远元, 王成泉, 王孟华, 杨洲鹏, 张红文, 王盛亮, 贾敬. 叠后反演技术在杨税务潜山裂缝孔隙型储层预测中的应用[J]. 物探与化探, 2020, 44(5): 1208-1214.
[7] 刘成功, 金胜, 魏文博, 景建恩, 叶高峰, 尹曜田. 高密度电阻率法比值参数基于阻尼最小二乘反演[J]. 物探与化探, 2019, 43(2): 351-358.
[8] 刘双, 刘天佑, 冯杰, 高文利, 邱礼泉. 蚁群算法在磁测资料反演解释中的应用[J]. 物探与化探, 2013, 37(1): 150-154.
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