摘要
针对智能变电站机器人在进行分闸作业时定位精度过低的问题,对基于单目视觉感知的定位方法进行了研究,并建立了单目视觉定位方案。首先针对变电站复杂的室内成像环境影响,设计了一种基于图像亮度均衡强化的阈值分割算法,可以从图像中准确提取开关的轮廓边缘信息;然后设计了一种基于RANSAC的椭圆拟合算法,对轮廓边缘信息进行拟合得到开关的位置信息;最后设计了基于灰度特征的快速匹配算法,并且融合了金字塔匹配策略,提高了匹配速度和抗干扰能力。实验结果表明,该方法的开关位置定位精度为±0.8mm,角度定位精度为±0.9°,且定位效率高,鲁棒性好,满足变电站机器人分闸作业的要求。
Aiming at the problem that the positioning accuracy of intelligent substation robot is too low when it is opening,the po⁃sitioning method based on monocular vision perception is studied,and a monocular vision positioning scheme is established.Firstly,a threshold segmentation algorithm based on image brightness equalization enhancement is designed for the complex in⁃door imaging environment of substation,which can accurately extract the contour edge information of the switch from the image;Then an ellipse fitting algorithm based on RANSAC is designed to fit the edge information of the contour to obtain the position in⁃formation of the switch;Finally,a fast matching algorithm based on gray features is designed,and the pyramid matching strat⁃egy is integrated to improve the matching speed and anti-interference ability.The experimental results show that the switch posi⁃tion positioning accuracy of this method is±0.8mm,the angle positioning accuracy is±0.9°,and the positioning efficiency is high and the robustness is good,meeting the requirements of the opening operation of the substation robot.
作者
方励云
周唯逸
朱星翰
王龙
FANG Li-yun;ZHOU Wei-yi;ZHU Xing-han;WANG Long(State Grid Zhejiang Yueqing Power Supply Co.,Ltd.,Zhejiang Wenzhou 325600,China;School of Mechanical and Electrical Engineering China Jiliang University,Zhejiang Hangzhou 310018,China)
出处
《机械设计与制造》
北大核心
2024年第9期266-270,共5页
Machinery Design & Manufacture
基金
国家重点研发计划项目(2018YFB2101004)。
关键词
单目视觉
定位识别
阈值分割
椭圆拟合
快速匹配
金字塔匹配
Monocular Vision
Location Recognition
Threshold Segmentation
Ellipse Fitting
Template Match⁃ing
Pyramidal Matching
作者简介
方励云,(1968-),男,浙江温州人,硕士研究生,高级工程师,主要研究方向:电网智能巡检技术。
