留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于多深度模型的钻孔结构面智能识别与量化分析

张野 陈金桥 李炎隆

张野, 陈金桥, 李炎隆. 基于多深度模型的钻孔结构面智能识别与量化分析[J]. 地质科技通报, 2023, 42(6): 31-41. doi: 10.19509/j.cnki.dzkq.tb20220091
引用本文: 张野, 陈金桥, 李炎隆. 基于多深度模型的钻孔结构面智能识别与量化分析[J]. 地质科技通报, 2023, 42(6): 31-41. doi: 10.19509/j.cnki.dzkq.tb20220091
Zhang Ye, Chen Jinqiao, Li Yanlong. Intelligent recognition and quantitative analysis of borehole hydraulic geological images utilizing multiple deep learning models[J]. Bulletin of Geological Science and Technology, 2023, 42(6): 31-41. doi: 10.19509/j.cnki.dzkq.tb20220091
Citation: Zhang Ye, Chen Jinqiao, Li Yanlong. Intelligent recognition and quantitative analysis of borehole hydraulic geological images utilizing multiple deep learning models[J]. Bulletin of Geological Science and Technology, 2023, 42(6): 31-41. doi: 10.19509/j.cnki.dzkq.tb20220091

基于多深度模型的钻孔结构面智能识别与量化分析

doi: 10.19509/j.cnki.dzkq.tb20220091
基金项目: 

国家自然科学基金项目 52009109

国家自然科学基金项目 52125904

博士启动基金项目 104-451120005

详细信息
    作者简介:

    张野(1990—), 男, 讲师, 主要从事水利工程安全评价和智能仿真研究工作。E-mail: zhangye1990@xaut.edu.cn

  • 中图分类号: TP391.41;P641.7

Intelligent recognition and quantitative analysis of borehole hydraulic geological images utilizing multiple deep learning models

  • 摘要:

    钻孔摄影图像是地质工程中分析深部岩体的有效方法和重要手段, 对于了解地表结构面延伸和地下结构面分布具有重要意义, 尤其对水利工程中复杂地质条件的分析有着不可或缺的作用。目前, 岩体中结构面的形态特征分析依赖于人工解译, 其结果具有一定的主观性, 同时不能保证解译的效率。为此提出了一种从数字钻孔摄像图像中智能分割结构面的方法, 并结合图像细化实现岩体结构面几何信息的量化分析。在研究中首先采用Unet、SegNet和DeepLabV3 3种不同的深度网络结构, 进行钻孔摄影图像结构面智能识别分析研究, 并将识别结果与传统的图像处理方法进行了比较。研究结果表明: 深度模型在复杂地质图像分割方面具有优越性。同时将Attention机制加入到编码-解码过程中, 提升了模型的准确性。在精确实现结构面分割的基础上, 采用图像细化方法提取出结构面的骨架, 对骨架图像进行单宽处理并提取有效信息, 最终实现结构面倾向、倾角和厚度的智能解译。对比结构面信息人工提取结果与自动识别结果, 倾向、倾角误差在3°以内, 厚度误差在0.65 mm以内。研究结果验证了本研究提出的结构面信息的智能量化方法的有效性, 在水利工程中具有广泛的应用前景。

     

  • 图 1  Unet模型结构

    Figure 1.  Unet architecture

    图 2  SegNet模型结构

    Figure 2.  SegNet architecture

    图 3  不同扩张率下的空洞卷积特征融合

    Figure 3.  Feature ensemble of dilated convolution with different expansion rates

    图 4  模型测试图像

    Figure 4.  Test images

    图 5  不同传统方法的图像分割结果

    Figure 5.  Segmentation of different traditional methods

    图 6  不同模型训练过程中IoULoss变化过程

    Figure 6.  IoU and Loss curves in training process of different models

    图 7  不同模型图像分割结果比较(1, 2, 3指区域编号)

    Figure 7.  Comparison of image segmentation results using different models

    图 8  测试图像分割结果

    Figure 8.  Test image segmentation results

    图 9  测试图像结构面骨架

    Figure 9.  Fracture skeleton of the test image

    图 10  结构面骨架去噪过程

    Figure 10.  Process of fracture skeleton noise removel

    表  1  不同模型性能评价比较

    Table  1.   Performance evaluation of different models

    训练模型 精确率/% 召回率/% F1_score/% IoU/%
    AUnet 99.2 99.2 99.2 83.1
    Unet 99.5 99.3 99.4 81.7
    ASegNet 99.0 98.9 98.9 79.5
    SegNet 98.7 98.7 98.6 76.4
    DeepLabV3 98.7 98.3 98.5 61.7
    下载: 导出CSV

    表  2  结构面信息手动提取和自动识别结果对比

    Table  2.   Comparison of manual and automatic fractures information calculations

    编号 倾向/(°) 倾角/(°) 厚度/mm
    人工提取 自动识别 差值 人工提取 自动识别 差值 人工提取 自动识别 差值
    I1-1 23.7 26.4 2.7 61.09 60.21 0.88 25.61 25.06 0.55
    I2-1 164.6 162.0 2.6 30.56 29.51 1.05 21.23 21.49 0.26
    I2-2 187.7 190.3 2.6 53.43 51.61 1.82 15.75 15.88 0.13
    I2-3 154.3 157.2 2.9 27.19 26.00 1.19 13.11 12.88 0.23
    I2-4 135.3 132.8 2.5 22.68 19.84 2.84 13.34 13.77 0.43
    I2-5 171.0 171.5 0.5 28.03 27.16 0.87 17.52 16.91 0.61
    I2-6 87.0 84.6 2.4 22.23 20.30 1.93 13.00 12.64 0.36
    I3-1 249.0 246.8 2.2 41.21 38.97 2.24 18.71 18.56 0.15
    I4-1 246.9 248.1 1.2 73.92 72.80 1.12 14.70 14.42 0.28
    I5-1 138.3 138.8 0.5 69.04 67.40 1.64 12.52 12.54 0.02
    I6-1 177.4 176.6 0.8 15.71 14.02 1.69 31.04 30.51 0.53
    I7-1 244.9 242.5 2.4 27.28 24.75 2.53 26.83 26.26 0.57
    I8-1 87.0 88.7 1.7 40.83 38.18 2.65 27.23 28.06 0.83
    I9-1 65.5 62.7 2.8 42.68 39.86 2.82 14.15 14.02 0.13
    I9-2 80.3 78.2 2.1 37.95 35.86 2.09 28.19 28.25 0.06
    I9-3 18.0 15.9 2.1 57.38 56.56 0.82 14.55 14.36 0.19
    I10-1 74.6 75.7 1.1 17.66 16.95 0.71 20.94 21.28 0.34
    下载: 导出CSV
  • [1] Schepers R, Rafat G, Gelbke C, et al. Application of borehole logging, core imaging and tomography to geotechnical exploration[J]. International Journal of Rock Mechanics and Mining Sciences, 2001, 38(6): 867-876. doi: 10.1016/S1365-1609(01)00052-1
    [2] 邹先坚, 王川婴, 韩增强, 等. 钻孔图像特征分析与结构面区域划分方法[J]. 天津大学学报: 自然科学与工程技术版, 2018, 51(1): 88-94.

    Zou X J, Wang C Y, Han Z Q, et al. Borehole image characteristic analysis and structural planes partitioning method[J]. Journal of Tianjin University: Science and Technology Edition, 2018, 51(1): 88-94(in Chinese with English abstract).
    [3] 季惠彬, 宋琨, 吕坤. 地下水封洞库深部岩体结构面连通性研究[J]. 地质科技情报, 2013, 32(1): 176-180. https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ201301033.htm

    Ji H B, Song K, Lü K. Connectivity of discontinuities of deep rock mass in water sealed underground storage cavern[J]. Geological Science and Technology Information, 2013, 32(1): 176-180(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DZKQ201301033.htm
    [4] 宣程强, 章杨松, 许文涛. 基于数字表面模型的岩体结构面产状获取[J]. 水文地质工程地质, 2022, 49(1): 75-83. doi: 10.16030/j.cnki.issn.1000-3665.202104029

    Xuan C Q, Zhang Y S, Xu W T. Extraction of the discontinuity orientation from a digital surface model[J]. Hydrogeology & Engineering Geology, 2022, 49(1): 75-83(in Chinese with English abstract). doi: 10.16030/j.cnki.issn.1000-3665.202104029
    [5] 葛云峰, 钟鹏, 唐辉明, 等. 基于钻孔图像的岩体结构面几何信息智能测量[J]. 岩土力学, 2019, 40(11): 4467-4476. doi: 10.16285/j.rsm.2018.1723

    Ge Y F, Zhong P, Tang H M, et al. Intelligent measurement on geometric information of rock discontinuities based on borehole image[J]. Rock and Soil Mechanics, 2019, 40(11): 4467-4476(in Chinese with English abstract). doi: 10.16285/j.rsm.2018.1723
    [6] 高金栋, 周立发, 冯乔, 等. 储层构造裂缝识别及预测研究进展[J]. 地质科技情报, 2018, 37(4): 158-166.

    Gao J D, Zhou L F, Feng Q, et al. Progress in reservior structural fracture characterization and prediction[J]. Geological Science and Technology Information, 2018, 37(4): 158-166(in Chinese with English abstract).
    [7] 徐志华, 郭戈, 孙钱程, 等. 改进的区域生长算法在三维激光点云识别岩体结构面中的应用[J/OL]. 水文地质工程地质, (2023-10-19)[2023-10-23]. http://kns.cnki.net/kcms/detail/11.2202.P.2202.P.20231018.1325.002.html.

    Xu Z H, Guo G, Sun Q C, et al. An improved region growing algorithm in 3D laser point cloud identification of rock mass structural plane[J/OL]. Hydrogeology & Engineering Geology, (2023-10-19)[2023-10-23]. http://kns.cnki.net/kcms/detail/11.2202.P.2202.P.20231018.1325.002.html(in Chinese with English abstract).
    [8] 王川婴, 胡培良, 孙卫春. 基于钻孔摄像技术的岩体完整性评价方法[J]. 岩土力学, 2010, 31(4): 1326-1330. doi: 10.16285/j.rsm.2010.04.008

    Wang C Y, Hu P L, Sun W C. Method for evaluating rock mass integrity based on borehole camera technology[J]. Rock and Soil Mechanics, 2010, 31(4): 1326-1330(in Chinese with English abstract). doi: 10.16285/j.rsm.2010.04.008
    [9] 黄达, 钟助. 基于单个钻孔孔壁电视图像确定地下岩体结构面产状的普适数学方法[J]. 地球科学: 中国地质大学学报, 2015, 40(6): 1101-1106. https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX201506015.htm

    Huang D, Zhong Z. A universal mathod for determining occurrence of underground rock discontinuity based on TV picture of wall of a single borehole[J]. Earth Science: Journal of China University of Geosciences, 2015, 40(6): 1101-1106(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX201506015.htm
    [10] Al-Sit W, Al-Nuaimy W, Marelli M, et al. Visual texture for automated characterisation of geological features in borehole televiewer imagery[J]. Journal of Applied Geophysics, 2015, 119: 139-146. doi: 10.1016/j.jappgeo.2015.05.015
    [11] Wang C, Zou X, Han Z, et al. An automatic recognition and parameter extraction method for structural planes in borehole image[J]. Journal of Applied Geophysics, 2016, 135(S1): 135-143.
    [12] 汪进超, 王川婴, 胡胜, 等. 孔壁钻孔图像的结构面参数提取方法研究[J]. 岩土力学, 2017, 38(10): 3074-3080.

    Wang J C, Wang C Y, Hu S, et al. A new method for extraction of structural surface in borehole images[J]. Rock and Soil Mechanics, 2017, 38(10): 3074-3080(in Chinese with English abstract).
    [13] 汪进超, 王川婴, 唐新建, 等. 基于钻孔摄像技术的岩体节理大小估算方法[J]. 岩土力学, 2017, 38(9): 2701-2707. https://www.cnki.com.cn/Article/CJFDTOTAL-YTLX201709033.htm

    Wang J C, Wang C Y, Tang X J, et al. A method for estimating rock mass joint size using borehole camera technique[J]. Rock and Soil Mechanics, 2017, 38(9): 2701-2707(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-YTLX201709033.htm
    [14] 李清波, 杜朋召. 基于边缘阈值分割的钻孔图像RQD自动分析方法研究[J]. 岩土工程学报, 2020, 42(11): 2153-2160. https://www.cnki.com.cn/Article/CJFDTOTAL-YTGC202011028.htm

    Li Q B, Du P Z. Automatic RQD analysis method based on information recognition of borehole images[J]. Chinese Journal of Geotechnical Engineering, 2020, 42(11): 2153-2160(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-YTGC202011028.htm
    [15] 宋琨, 孙驰, 安冬, 等. 数字钻孔全景影像中结构面特征智能识别方法[J]. 地质科技通报, 2020, 39(5): 17-22. doi: 10.19509/j.cnki.dzkq.2020.0503

    Song K, Sun C, An D, et al. Intelligent identification method for rock discontinuities properties by digital borehole panoramic images[J]. Bulletin of Geological Science and Technology, 2020, 39(5): 17-22(in Chinese with English abstract). doi: 10.19509/j.cnki.dzkq.2020.0503
    [16] Sapoval N, Aghazadeh A, Nute M G, et al. Current progress and open challenges for applying deep learning across the biosciences[J]. Nature Communications, 2022, 13(1): 1728.
    [17] Choudhary K, DeCost B, Chen C, et al. Recent advances and applications of deep learning methods in materials science[J]. NPJ Computational Materials, 2022, 8(1): 59.
    [18] 刘彦锋, 张文彪, 段太忠, 等. 深度学习油气藏地质建模研究进展[J]. 地质科技通报, 2021, 40(4): 235-241. doi: 10.19509/j.cnki.dzkq.2021.0417

    Liu Y F, Zhang W B, Duan T Z, et al. Progress of deep learningin oil and gas reservoir geological modeling[J]. Bulletin of Geological Science and Technology, 2021, 40(4): 235-241(in Chinese with English abstract). doi: 10.19509/j.cnki.dzkq.2021.0417
    [19] 李明超, 田丹, 沈扬, 等. 融入Attention机制改进Word2vec技术的水利水电工程专业词智能提取与分析方法[J]. 水利学报, 2020, 51(7): 816-826.

    Li M C, Tian D, Shen Y, et al. An intelligent extraction and analysis approach of professional technical words for hydraulic engineering by improved Word2vec technology with Attention mechanism[J]. Journal of Hydraulic Engineering, 2020, 51(7): 816-826(in Chinese with English abstract).
    [20] 李明超, 符家科, 张野, 等. 耦合岩石图像与锤击音频的岩性分类智能识别分析方法[J]. 岩石力学与工程学报, 2020, 39(5): 996-1004. https://www.cnki.com.cn/Article/CJFDTOTAL-YSLX202005013.htm

    Li M C, Fu J K, Zhang Y, et al. Intelligent recognition and analysis method of rock lithology classification based on coupled rock images and hammering audios[J]. Chinese Journal of Rock Mechanics and Engineering, 2020, 39(5): 996-1004(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-YSLX202005013.htm
    [21] Li Y. Research and application of deep learning in image recognition[C]//Anon. 2022 IEEE 2nd International Conference on Power, Electronics and Computer Applications (ICPECA). [S. l. ]: IEEE, 2022: 994-999.
    [22] Chen J J, Lu W S, Yuan L, et al. Estimating construction waste truck payload volume using monocular vision[J]. Resources, Conservation and Recycling, 2022, 177: 106013.
    [23] Huang H, Li Q, Zhang D. Deep learning based image recognition for crack and leakage defects of metro shield tunnel[J]. Tunnelling and Underground Space Technology, 2018, 77: 166-176.
    [24] Liang X. Image-based post-disaster inspection of reinforced concrete bridge systems using deep learning with Bayesian optimization[J]. Computer-Aided Civil and Infrastructure Engineering, 2019, 34(5): 415-430.
    [25] 陈波, 张华, 汪双, 等. 基于全卷积神经网络的坝面裂纹检测方法研究[J]. 水力发电学报, 2020, 39(7): 52-60.

    Chen B, Zhang H, Wang S, et al. Study on detection method of dam surface cracks based on full convolution neural network[J]. Journal of Hydroelectric Engineering, 2020, 39(7): 52-60(in Chinese with English abstract).
    [26] 雷雨萌, 陈祖煜, 于沭, 等. 基于深度阈值卷积模型的土石料级配智能检测方法研究[J]. 水利学报, 2021, 52(3): 369-380. https://www.cnki.com.cn/Article/CJFDTOTAL-SLXB202103013.htm

    Lei Y M, Chen Z Y, Yu S, et al. Intelligent detection for earth-rockfill materials base on Deep Otsu Convolutional Neural Network[J]. Journal of Hydraulic Engineering, 2021, 52(3): 369-380(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-SLXB202103013.htm
    [27] 薛东杰, 唐麒淳, 王傲, 等. 基于FCN的岩石混凝土裂隙几何智能识别[J]. 岩石力学与工程学报, 2019, 38(增刊2): 3393-3403.

    Xue D J, Tang Q C, Wang A, et al. FCN-based intelligent identification of crack geometry in rock or concrete[J]. Chinese Journal of Rock Mechanics and Engineering, 2019, 38(S2): 3393-3403(in Chinese with English abstract).
    [28] 王超, 贾贺, 张社荣, 等. 基于图像的混凝土表面裂缝量化高效识别方法[J]. 水力发电学报, 2021, 40(3): 134-144. https://www.cnki.com.cn/Article/CJFDTOTAL-SFXB202103014.htm

    Wang C, Jia H, Zhang S R, et al. Image-based quantitative and efficient identification method for concrete surface cracks[J]. Journal of Hydroelectric Engineering, 2021, 40(3): 134-144(in Chinese with English abstract). https://www.cnki.com.cn/Article/CJFDTOTAL-SFXB202103014.htm
    [29] 任秋兵, 李明超, 沈扬, 等. 水工混凝土裂缝像素级形态分割与特征量化方法[J]. 水力发电学报, 2021, 40(2): 234-246.

    Ren Q B, Li M C, Shen Y, et al. Pixel-level shape segmentation and feature quantification of hydraulic concrete cracks based on digital images[J]. Journal of Hydroelectric Engineering, 2021, 40(2): 234-246(in Chinese with English abstract).
    [30] Karimpouli S, Tahmasebi P. Segmentation of digital rock images using deep convolutional autoencoder networks[J]. Computers & Geosciences, 2019, 126: 142-150.
    [31] König J, Jenkins M D, Mannion M, et al. Optimized deep encoder-decoder methods for crack segmentation[J]. Digital Signal Processing, 2020, 108: 102907.
    [32] Otsu N. A threshold selection method from gray-level histograms[J]. IEEE Transactions on Systems, Man and Cybernetics, 1979, 9(1): 62-66.
    [33] Ronneberger O, Fischer P, Brox T. Unet: Convolutional networks for biomedical image segmentation[C]//Anon. Proceedings of International Conference on Medical Image Computing and Computer-Assisted Intervention. Berlin: Springer, 2015: 234-241.
    [34] Badrinarayanan V, Kendall A, Cipolla R. SegNet: A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(12): 2481-2495.
    [35] Chen L C, Papandreou G, Schroff F, et al. Rethinking atrous convolution for semantic image segmentation[J]. arXiv preprint arXiv, 2017: 1706.05587.
    [36] Xu J J, Zhang H, Tang C S, et al. Automatic soil crack recognition under uneven illumination condition with the application of artificial intelligence[J]. Engineering Geology, 2022, 296: 106495.
    [37] Cui X, Wang Q, Dai J, et al. Intelligent crack detection based on attention mechanism in convolution neural network[J]. Advances in Structural Engineering, 2021, 24(9): 1859-1868. http://www.researchgate.net/publication/348599780_Intelligent_crack_detection_based_on_attention_mechanism_in_convolution_neural_network
    [38] Kang J, Liu L, Zhang F, et al. Semantic segmentation model of cotton roots in-situ image based on attention mechanism[J]. Computers and Electronics in Agriculture, 2021, 189: 106370.
    [39] Zhao X, Wang S, Zhao J, et al. Application of an attention U-Net incorporating transfer learning for optic disc and cup segmentation[J]. Signal, Image and Video Processing, 2021, 15(5): 913-921. doi: 10.1007/s11760-020-01815-z
  • 加载中
图(10) / 表(2)
计量
  • 文章访问数:  358
  • PDF下载量:  48
  • 被引次数: 0
出版历程
  • 收稿日期:  2022-03-07
  • 录用日期:  2022-05-11
  • 修回日期:  2022-05-05

目录

    /

    返回文章
    返回