基于回归模型的电力线路三维全景监测方法  

Three-dimensional Panoramic Monitoring Method of Power Line Based on Regression Model

作　　者：陈赟 朱超杰 周亮 CHEN Yun;ZHU Chaojie;ZHOU Liang(Economic Research Institute,State Grid Shanghai Electric Power Company,Shanghai 200002,China)

机构地区：[1]国网上海市电力公司经济技术研究院,上海200002

出　　处：《机械制造与自动化》2023年第6期142-146,共5页Machine Building & Automation

摘　　要：针对电力线路监控过程监控与真实场景有割裂感、监测数据来源单一导致三维全景电力线路监测存在清晰度低、异常报警率低等问题,设计一种基于回归模型的电力线路三维全景监测方法。通过DM6446双核控制芯片,在ARM端搭载实时操作系统,实现电力线路图像的采集、管理以及显示等功能。构建多源异构数据的协同特征回归模型,提取电力线路图像特征,结合信息共享方法,获取元信息和残差信息,通过不同的统计量完成实时处理和电力线路三维全景监测。实验结果表明:三维全景监测方法的特征提取数量达到了532个,可以清晰呈现电力线路三维全景,异常报警率为97.3%。To cope with the low definition and low abnormal alarm rate in three-dimensional panoramic power line monitoring due to the separation of power line monitoring process from real scene and unitary monitoring data source,a three-dimensional panoramic power line monitoring method based on regression model is proposed.With the DM6446 dual core control chip,the ARM terminal is equipped with a real-time operating system to realize the collection,management,display and other related functions of power line images.The collaborative feature regression model of multi-source heterogeneous data is constructed to extract the power line features.Combined with the information sharing method,the meta-information and residual information is abtained.The real-time 3D panoramic monitoring is completed through different statistics to realize the 3D panoramic monitoring of power lines.The experimental results show that the number of feature extraction of the three-dimensional panoramic monitoring method reaches 532,which clearly presents the three-dimensional panorama of power lines with the abnormal alarm rate being 97.3%.

关 键 词：回归模型 电力线路 三维全景 特征提取 图像监测 特征点提取 异常报警 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]