基于FCN的岩石混凝土裂隙几何智能识别

doi:10.13722/j.cnki.jrme.2019.0010

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (1092 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要裂隙几何的精准、高效识别是建立岩石混凝土材料非连续力学模型的基础。传统算法，如阈值分割、边缘检测与区域生长法等，基于有限经验，工作量大且识别精度低。人工智能识别迥异于人类经验，可自我循环、可自我正反馈甚至可实现自我学习。人造类材料中混凝土性能最接近于岩石，具有相似的裂纹几何分布特征。针对天然岩石与混凝土裂隙识别，尝试提出基于全卷积神经网络的智能识别算法。首先建立基于混凝土3种应用场景(建筑结构、路面及隧道表面)的裂隙数据集，经过卷积操作、池化操作与反卷积操作实现了误差值的迅速收敛。引入新的卷积核参数，架构新的全卷积网络，基于像素的二分类问题，利用查准率与查全率对识别模型进行评价，全卷积神经网络识别算法评价指标较好。进而结合矢量化算法，实现了裂隙几何长度、宽度及面积的实时统计。最后对边缘检测、大律法、区域生长法及全卷积神经网络算法进行了验证，任意选取5组图像进行识别，新算法提取的裂隙与标签图在定性上最相近，并且在定量上查全率与查准率也极大优于传统算法。全卷积神经网络随着自我学习的深入还可不断提高识别精度，降低误差，未来在岩石、混凝土等工程领域的应用具有巨大生命力。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	薛东杰1
	2
	3
	4
	唐麒淳1
	王傲5
	张辽1
	周宏伟2
	5

关键词 ：岩石力学, 裂隙识别, 传统算法, 人工智能识别, 全卷积神经网络, 查准率, 查全率

Abstract：Efficient and accurate extraction of crack geometry is fundamental for rock or concrete engineering. Traditional algorithms of crack identification，such as threshold segmentation，edge detection and region growing method，are based on limited human experience with tedious work and low accuracy. Artificial intelligence(AI) recognition is self-circulating dependent on big data，and can be positive feedback or even self-learning. The concrete as one of many artificial materials has the similar properties with rock，especially in geometric distribution. An intelligent recognition algorithm based on full convolutional neural network is proposed to identify cracks on rock and concrete surface. The dataset of crack geometry commonly seen in three scenarios of concrete used in building structure，pavement and tunnel surface is established. Through convolution operation，pooling operation and deconvolution operation，the error is quickly reduced with training times. A new convolution kernel parameter is introduced into a new full convolutional network，and a recognition model is evaluated based on the pixel-based two-classification problem. The results shows the better recognition of the new full convolutional network than traditional algorithms. Combined with the vectorization algorithm，the real-time statistical analysis of crack geometry including length，width and area is realized. Finally，the crack identification determined by artificial intelligence and traditional algorithm is compared by arbitrarily selecting five groups of images. The cracks extracted by the new algorithm are qualitatively similar to the ground truth，and the precision and recall are higher than the traditional algorithms. The full convolutional network can continuously improve the recognition accuracy，reduce the error，and will show great vitality in the application of rock or concrete engineering.

Key words： rock mechanics crack identification traditional algorithm artificial intelligence identification fully convolutional network(FCN) precision recall

引用本文:

薛东杰1，2，3，4，唐麒淳1，王傲5，张辽1，周宏伟2，5. 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(S2): 3393-3403.
XUE Dongjie1，2，3，4，TANG Qichun1，WANG Ao5，ZHANG Liao1，ZHOU Hongwei2，5. FCN-based intelligent identification of crack geometry in rock or concrete. , 2019, 38(S2): 3393-3403.

链接本文:

https://rockmech.whrsm.ac.cn/CN/10.13722/j.cnki.jrme.2019.0010 或 https://rockmech.whrsm.ac.cn/CN/Y2019/V38/IS2/3393

[1]	SILVER D，HUANG A，MADDISON C J，et al. Mastering the game of Go with deep neural networks and tree search[J]. Nature，2016，529：484–489.
[2]	BARTON N，LIEN R，LUNDE J. Engineering classification of rock masses for the design of tunnel support[J]. Rock Mechanics，1974，6(4)：189–236.
[3]	JAEGER J C，COOK N G W，ZIMMERMAN R. Fundamentals of rock mechanics[M]. [S. l.]：John Wiley and Sons，2009：1–8.
[4]	AZARAFZA M，GHAZIFARD A，AKGüN H，et al. Development of a 2D and 3D computational algorithm for discontinuity structural geometry identification by artificial intelligence based on image processing techniques[J]. Bulletin of Engineering Geology and the Environment，2019：78(5)：3 371–3 383.
[5]	NASIRI S，KHOSRAVANI M R，WEINBERG K. Fracture mechanics and mechanical fault detection by artificial intelligence methods：a review[J]. Engineering Failure Analysis，2017，81：270–293.
[6]	HINTON G E，SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science，2006，313(5786)：504–507.
[7]	ELKATATNY S，TARIQ Z，MAHMOUD M，et al. An integrated approach for estimating static Young?s modulus using artificial intelligence tools[J]. Neural Computing and Applications，2019，31(8)：4 123–4 133.
[8]	ELKATATNY S. Application of artificial intelligence techniques to estimate the static Poisson's ratio based on wireline Log data[J]. Journal of Energy Resources Technology，2018，140(7)：1–8.
[9]	SONMEZ H，GOKCEOGLU C，NEFESLIOGLU H A，et al. Estimation of rock modulus：for intact rocks with an artificial neural network and for rock masses with a new empirical equation[J]. International Journal of Rock Mechanics and Mining Sciences，2006，43(2)：224–235.
[10]	MAJDI A，BEIKI M. Evolving neural network using a genetic algorithm for predicting the deformation modulus of rock masses[J]. International Journal of Rock Mechanics and Mining Sciences，2010，47(2)：246–253.
[11]	DREUZY J R，MéHEUST Y，PICHOT G. Influence of fracture scale heterogeneity on the flow properties of three-dimensional discrete fracture networks(DFN)[J]. Journal of Geophysical Research：Solid Earth，2012，117(B11)：1–21.
[12]	HUI M H，MALLISON B T，FYROZJAEE M H，et al. The upscaling of discrete fracture models for faster，coarse-scale simulations of IOR and EOR processes for fractured reservoirs[C]// SPE Annual Technical Conference and Exhibition. Society of Petroleum Engineers. [S. l.]：[s. n.]，2013：1–15.
[13]	KRANZ R L. Microcracks in rocks：a review[J]. Tectonophysics，1983，100(1/3)：449–480.
[14]	WONG T，WONG R H C，CHAU K T，et al. Microcrack statistics，Weibull distribution and micromechanical modeling of compressive failure in rock[J]. Mechanics of Materials，2006，38(7)：664–681.
[15]	HARTHONG B，SCHOLTèS L，DONZé F V. Strength characterization of rock masses，using a coupled DEM-DFN model[J]. Geophysical Journal International，2012，191(2)：467–480.
[16]	王睿，漆泰岳，雷波，等. 隧道衬砌裂隙特征提取方法研究[J]. 岩石力学与工程学报，2015，34(6)：1 211–1 217.(WANG Rui，QI Taiyue，LEI Bo，et al. Study on the characteristic extraction of tunnel lining cracks[J]. Chinese Journal of Rock Mechanics and Engineering，2015，34(6)：1 211–1 217.(in Chinese))
[17]	王华，朱宁，王祁. 应用计盒维数方法的路面裂隙图像分割[J]. 哈尔滨工业大学学报，2007，(1)：142–144.(WANG Hua，ZHU Ning，WANG Qi. Segmentation of pavement cracks using differential box-counting approach[J]. Journal of Harbin Institute of Technology，2007，(1)：142–144.(in Chinese))
[18]	刘凡凡，徐国爱，肖靖，等. 基于连通域相关及Hough变换的公路路面裂隙提取[J]. 北京邮电大学学报，2009，32(2)：24–28. (LIU Fanfan，XU Guoai，XIAO Jing，et al. Cracking automatic extraction of pavement based on connected domain correlating and hough transform[J]. Journal of Beijing University of Posts and Telecommunications，2009，32(2)：24–28.(in Chinese))
[19]	王世芳，车艳丽，李楠，等. 一种基于多尺度脊边缘的沥青路面裂隙检测算法[J]. 中国公路学报，2017，30(4)：32–41.(WANG Shifang，CHE Yanli，LI Nan，et al. Asphalt pavement crack detection algorithm based on multi-scale ridges[J]. China Journal of Highway and Transport，2017，30(4)：32–41.(in Chinese))
[20]	许薛军，张肖宁. 基于数字图像的混凝土桥梁裂隙检测技术[J]. 湖南大学学报：自然科学版，2013，40(7)：34–40.(XU Xuejun，ZHANG Xiaoning. Crack detection of concrete bridges based digital image[J]. Journal of Hunan University：Natural Sciences，2013，40(7)：34–40.(in Chinese))
[21]	苟量，彭真明. 小波多尺度边缘检测及其在裂隙预测中的应用[J]. 石油地球物理勘探，2005，(3)：309–313.(GOU Liang，PENG Zhenming. Multi-scale edge detection of wavelet and application in fracture prediction[J]. Oil Geophysical Prospecting，2005，(3)：309–313.(in Chinese))
[22]	朱平哲，黎蔚. 基于主动生长的断裂裂隙块的连接方法[J]. 计算机应用，2011，31(12)：3 382–3 384.(ZHU Pingzhe，LI Wei. Linking algorithm of discontinuity crack block based on autonomous edge growing[J]. Journal of Computer Applications，2011，31(12)：3 382–3 384.(in Chinese))
[23]	闫东阳，明冬萍. 基于自动多种子区域生长的遥感影像面向对象分割方法[J]. 工程科学学报，2017，39(11)：1 735–1 742.(YAN Dongyang，MING Dongping. Object-oriented remote sensing image segmentation based on automatic multiseed region growing algorithm[J]. Chinese Journal of Engineering，2017，39(11)：1 735–1 742.(in Chinese))
[24]	肖明尧，李雄飞，张小利，等. 基于多尺度的区域生长的图像分割算法[J]. 吉林大学学报：工学版，2017，47(5)：1 591–1 597.(XIAO Mingxiao，LI Xiongfei，ZHANG Xiaoli，et al. Medical image segmentation algorithm based on multi-scale region growing[J]. Journal of Jilin University：Engineering and Technology，2017，47(5)：1 591–1 597.(in Chinese))
[25]	崔冰，王卫星. 基于统计模式识别的岩石节理图像分割方法[J]. 计算机工程与应用，2006，(10)：213–215.(CUI Bing，WANG Weixing. Statistical pattern recognition for segmentation of rock joint image[J]. Computer Engineering and Applications，2006，(10)：213–215.(in Chinese))
[26]	陈立万. 遗传优化BP神经网络在岩石节理图像分割中的应用[J]. 微计算机信息，2010，26(23)：211–213.(CHEN Liwan. Application of genetic optimizing BPNN to rock fracture image segmentation[J]. Microcomputer Information，2010，26(23)：211–213.(in Chinese))
[27]	陶开云，魏海，廖敏，等. 一种基于SVM的坝面裂隙损伤智能识别方法[J]. 价值工程，2017，36(19)：124–125.(TAO Kaiyun，WEI Hai，LIAO Min，et al. A dam surface crack damage scale intelligent recognition based on SVM method[J]. Value Engineering，2017，36(19)：124–125.(in Chinese))
[28]	CHA Y J，CHOI W，BüYüK?ZTüRK，et al. Deep learning-based crack damage detection using convolutional neural networks[J]. Computer-Aided Civil and Infrastructure Engineering，2017，32(5)：361–378.
[29]	ZHANG L，YANG F，ZHANG Y，et al. Road crack detection using deep convolutional neural network[C]// Image Processing(ICIP)，2016 IEEE International Conference on. [S. l]：[s. n.]，2016：3 708–3 712.
[30]	赵珊珊，何宁. 基于卷积神经网络的路面裂隙检测[J]. 传感器与微系统，2017，36(11)：135–138.(ZHAO Shanshan，HE Ning. Road surface crack detection based on CNN[J]. Transducer and Microsystem Technologies，2017，36(11)：135–138.(in Chinese))
[31]	KURT H，MAXWELL S，HALBERT W. Multilayer feedforward networks are universal approximators[J]. Neural network，1989，2(5)：359–366.
[32]	周飞燕，金林鹏，董军. 卷积神经网络研究综述[J]. 计算机学报，2017，40(6)：1 229–1 251.(ZHOU Feiyan，JIN Linpeng，DONG Jun. Review of convolutional neural network[J]. Chinese Journal of Computers，2017，40(6)：1 229–1 251.(in Chinese))
[33]	黄宏伟，李庆桐. 基于深度学习的盾构隧道渗漏水病害图像识别[J]. 岩石力学与工程学报，2017，36(12)：2 861–2 871.(HUANG Hongwei，LI Qingtong. Image recognition for water leakage in shield tunnel based on deep learning[J]. Chinese Journal of Rock Mechanics and Engineering，2017，36(12)：2 861–2 871.(in Chinese))
[34]	薛亚东，李宜城. 基于深度学习的盾构隧道衬砌病害识别方法[J].湖南大学学报：自然科学版，2018，45(3)：100–109.(XUE Yadong，LI Yicheng. A method of disease recognition for shield tunnel lining based on deep learning[J]. Journal of Hunan University：Natural Sciences，2018，45(3)：100–109.(in Chinese))
[35]	苑玮琦，薛丹. 基于机器视觉的隧道衬砌裂隙检测算法综述[J]. 仪器仪表学报，2017，38(12)：3 100–3 111.(YUAN Weiqi，XUE Dan. Review of tunnel lining crack detection algorithm based on machine vision[J]. Chinese Journal of Scientific Instrument，2017，38(12)：3 100–3 111(in Chinese))
[36]	朱力强，白彪，王耀东，等. 基于特征分析的地铁隧道裂隙识别算法[J]. 铁道学报，2015，37(5)：64–70.(ZHU Liqiang，BAI Biao，WANG Yaodong，et al. Subway tunnel crack identification algorithm based on feature analysis[J]. Journal of the China Railway Society，2015，37(5)：64–70.(in Chinese))
[37]	SELINGER P. Potrace：a polygon-based tracing algorithm[J/OL]. http：//potrace. sourceforge. net/potrace. pdf，2003.