基于三维智能路面检测系统识别沉陷病害的方法研究  

Based on the Research of 3D Intelligent Pavement Detection System to Identify the Disease of Subsidence

作　　者：陈梦月 孔海望 吴迪 CHEN Mengyue;KONG Haiwang;WU Di(Guangdong Jianke Traffic Engineering Quality Inspection Center Co.,Ltd.Guangzhou 510765,China)

机构地区：[1]广东建科交通工程质量检测中心有限公司,广州510765

出　　处：《广东土木与建筑》2022年第5期15-18,共4页Guangdong Architecture Civil Engineering

摘　　要：为提高沉陷病害检出率从而保障道路运营安全,研究了针对沉陷病害的识别方法。根据实际情况,与人工识别沉陷方法进行对比,分别分析了坑槽、井盖、车辙对检测结果的影响。利用三角位移和激光测距原理得到路面纵断面高程曲线,先从高程曲线中筛选出-10 mm以下范围数据,然后剔除通过基于神经网络算法的人工智能技术识别出的坑槽、井盖以及通过多维度特征参数提取的车辙路段,剩余-10 mm以下范围路段数据即为沉陷病害,试验结果表明:该识别路面沉陷病害的方法较好,效率高,准确性好,对轻、重度沉陷的识别准确率分别达到70%、80%。In order to improve the detection rate of subsidence disease and ensure the safety of road operation,the identification method of subsidence disease is studied.According to the actual situation,compared with the manual method to identify the subsidence,the influence of pit slot,manhole cover and rut on the detection result is analyzed.Based on the principle of triangular displacement and laser ranging,the elevation curve of longitudinal section of road surface is obtained,then,the pits,manholes and rutting sections identified by artificial intelligence technology based on neural network algorithm and extracted by multi-dimensional characteristic parameters are eliminated,and the remaining road section data below-10 mm is the subsidence disease,the test results show that the method is effective,efficient and accurate,and the accuracy of light and heavy subsidence can reach 70%and 80%respectively.

关 键 词：纵断面高程 三角位移 激光测距 神经网络算法 多维度特征 

分 类 号：U418.6[交通运输工程—道路与铁道工程]