大坝边坡测斜孔变形自动化监测及变形模式识别研究  

Study on Automatic Monitoring of Deformation in Slope Inclinometer Holes and Deformation Pattern Recognition of Dams

作　　者：冉鲁光 周小燕[1] 李双平 张斌 刘祖强 王华为 李建川 RAN Lu-guang;ZHOU Xiao-yan;LI Shuang-ping;ZHANG Bin;LIU Zu-qiang;WANG Hua-wei;LI Jian-chuan(Xiangjiaba Power Plant of China Yangtze Power Co.,Ltd.,Yibin 644612,China;Changjiang Spatial Information Technology Engineering Co.,Ltd.,Wuhan 430010,China)

机构地区：[1]中国长江电力股份有限公司向家坝电厂,四川宜宾644612 [2]长江空间信息技术工程有限公司(武汉),湖北武汉430010

出　　处：《水电能源科学》2025年第3期147-151,共5页Water Resources and Power

基　　金：中国长江电力股份有限公司科研项目(Z422302005)。

摘　　要：针对大坝边坡深部位移监测的自动化需求,开发了一种基于物联网的钻孔测斜机器人系统,并创新性地结合一维卷积神经网络模型(1D CNN)实现边坡变形模式的智能预测。通过自主研发的硬件主控板,实现了测斜仪的自动化控制及数据的实时采集和传输。基于采集的深部位移数据,1D CNN模型自动提取曲线特征并进行分类,识别出多种变形模式(如变形稳定、剪切滑动等),从而对边坡变形趋势进行智能化预测,有效支持地质灾害预警。试验表明,测斜机器人在A、B向的测量精度分别达±0.82、±1.04 mm/30 m,且1D CNN模型在曲线分类上表现优异。该系统通过高精度监测与自动化分析,显著提升了大坝边坡的监测效率和预警水平,具备广泛应用的潜力。This study addresses the need for automated monitoring of deep displacement in dam slopes by developing an IoT-based borehole inclinometer robot system,incorporating an innovative one-dimensional convolutional neural network model(1D CNN)for intelligent prediction of slope deformation patterns.This system uses a self-developed hardware control board to enable automated inclinometer control and real-time data acquisition and transmission.Based on the collected deep displacement data,the 1D CNN model automatically extracts curve features and classifies them,identifying various deformation modes(e.g.,stable deformation,shear sliding)and enabling intelligent prediction of slope deformation trends to support geohazard early warning.Experimental results show that the inclinometer robot achieves measurement accuracy of±0.82 mm/30 m in the A direction and±1.04 mm/30 m in the B direction,with the 1D CNN model demonstrating excellent performance in curve classification.Through high-precision monitoring and automated analysis,this system significantly enhances monitoring efficiency and early warning capabilities for dam slopes,offering extensive application potential.

关 键 词：测斜机器人 物联网 变形模式 卷积神经网络 大坝边坡 

分 类 号：TV698.1[水利工程—水利水电工程]