基于LiDAR点云的电力线自适应密度聚类提取

Adaptive density clustering for extracting power line based on LiDAR point clouds

目前,高压电力线巡检效率已不能满足新时期电力系统智能化管理的要求。通过深入探讨密度聚类法在机载LiDAR电力线点云提取中存在的弊端,给出自适应—密度聚类解决方案。实测青海省某电力走廊机载LiDAR点云数据,借助Visual Studio 2010 C++开发环境,编制自适应—密度聚类等相关处理程序,对自适应密度聚类方案的电力线点云提取、电力线三维抛物线的拟合进行测试与精度评定,结果表明:①自适应—密度聚类方案提取准确率达99.96%,电力线拟合最小残差0.220m,最大拟合残差0.252m,平均拟合残差0.232m;②自适应—密度聚类方案一次便可成功提取电力线,较好地规避了密度聚类法中多次试探邻域半径r_(Eps)与密度阈值p_(MinPts)等初始参数的赋值问题,大大提高了基于机载LiDAR点云数据的电力巡线工作效率,可应用于电力行业的实际工作中。

At present,the inspection efficiency of high-voltage power line cannot meet the requirements of intelligent management of power system in the new era.By deeply analyzing the disadvantages of density clustering method in power line extraction with airborne LiDAR point clouds,an adaptive density clustering solution is given.Based on the measured airborne LiDAR point clouds of a power corridor in Qinghai Province,with the help of Visual Studio 2010 C++development environment,relevant processing programs such as adaptive density clustering are compiled to test and evaluate the accuracy of power line extraction and power line three-dimensional parabola fitting.The results show that:The extraction accuracy of the adaptive density clustering scheme is 99.96%,the minimum residual of power line fitting is 0.220m,the maximum is 0.252m,and the average is 0.232m;②The adaptive density clustering scheme successfully extracts the power line at one time,which avoids the assignment of initial parameters such as neighborhood radius and density threshold in the density clustering method for many times,greatly improves the work efficiency,and can be applied to the actual work of power industry.