基于机器学习的隧道围岩岩性智能识别方法  

Intelligent Lithology Identification Method of Tunnel Surrounding Rock Based on Machine Learning

作　　者：吴全德 马治中 郭珂依 刘大刚[1] 高洪涛 左继功 WU Quande;MA Zhizhong;GUO Keyi;LIU Dagang;GAO Hongtao;ZUO Jigong(School of Civil Engineering,Southwest Jiaotong University,Chengdu 610031,China;The 3rd Engineering Co.,Ltd.of China Railway Construction Bridge Engineering Bureau Group,Shenyang 110043,China;Xi'an—Chengdu High-speed Railway Shaanxi Co.,Ltd.,Xi'an 710000,China)

机构地区：[1]西南交通大学土木工程学院,成都610031 [2]中铁建大桥工程局集团第三工程有限公司,沈阳110043 [3]西成铁路客运专线陕西有限责任公司,西安710000

出　　处：《路基工程》2024年第6期49-54,共6页Subgrade Engineering

基　　金：中国国家铁路集团有限公司系统性重大项目:铁路隧道机械化施工关键技术研究(P2018G048)。

摘　　要：为实现智能高效且可靠的隧道围岩岩性识别分类,采用k近邻、支持向量机、随机森林和梯度提升树4种机器学习算法,对砂岩、灰岩、花岗岩和片麻岩进行岩性识别研究。采用铜川隧道掌子面围岩及网络岩石图像进行测试,通过提取岩石图像H、S、V均值,构建岩性识别特征空间,结合机器学习算法原理,建立特征空间与岩石类别间的映射关系,以识别准确率和运行时间为评价指标,对比分析4种算法的识别效果。结果表明:k近邻、随机森林和梯度提升树均具有较高的识别准确率,综合考虑算法精度与算法效率,建议将k近邻算法作为优选算法。In order to realize intelligent,efficient and reliable lithology identification and classification of tunnel surrounding rock,four machine learning algorithms,including k-nearest neighbor,support vector machine,random forest and gradient lifting tree,are used to identify lithology of sandstone,limestone,granite and gneiss.The surrounding rock of the face of the Tongchuan tunnel and the network rock images were used for testing.The average H,S and V values of the rock images were extracted to construct the lithology identification feature space.Based on the principle of machine learning algorithm,the mapping relationship between the feature space and the rock category is established,and the recognition accuracy and running time are taken as evaluation indexes to compare and analyze the recognition effects of the four algorithms.The results show that k-nearest neighbor,random forest and gradient lifting tree all have high recognition accuracy.Considering the algorithm accuracy and efficiency,it is suggested that k-nearest neighbor algorithm should be used as the optimal algorithm.

关 键 词：隧道工程 机器学习 岩性识别 颜色空间 岩石图像 

分 类 号：U25[交通运输工程—道路与铁道工程]