| With the rapid construction of power lines in China,the inspection workload of power workers has also increased significantly.The manual inspection method has long been unable to meet the demand of the current workload.At present,the inspection method combining manual and UAV has also begun to be stretched.Using UAV to carry out independent inspection of power poles and towers can greatly improve work efficiency,reduce power grid maintenance cost and liberate productivity,It has important research significance and broad application prospects.This paper studies the UAV autonomous inspection technology of power tower,and focuses on the accurate positioning method to eliminate the position error caused by GNSS navigation when UAV performs the inspection task.Starting from the characteristics of inspecting power tower and the key technologies used in it,the UAV inspection path is planned,and the architecture of UAV autonomous inspection system of power tower is designed,including UAV inspection flight platform and data processing system.On this basis,the GNSS positioning test of UAV is carried out,and its positioning performance and error reasons are analyzed.From the perspective of image feature matching,pixel motion and deep learning target detection,three image positioning methods based on visual f eedback are proposed to eliminate GNSS positioning error.Firstly,a bag of words model image matching and location method based on ORB feature is proposed.By extracting the ORB features of the image and constructing the visual dictionary,the image query and feature matching are carried out on this basis,and the pose transformation matrix between the images is solved to realize the position adjustment of the UAV.Secondly,an image location method based on sparse direct method is proposed.Through the order of inspection points,select the standard image at the inspection point,track the pixel motion on the standard image and UAV positioning image according to the assumption of constant gray level,and solve the pose transformation matrix between the two images to guide the position adjustment of UAV.Thirdly,an image location method based on YOLOv5 target detection is proposed.The relative position of the insulator on the UAV frame is adjusted according to the image training of the UAV,and the relativ e position of the insulator on the UAV frame is recognized according to the image.Finally,it is determined that the image positioning method based on YOLOv5 target detection can achieve good results.The power line simulation environment is built in the gazebo simulation platform,the visual position controller of UAV is designed,and the UAV autonomous inspection test of power tower is carried out.The experimental results show that this method can eliminate the G NSS positioning error of UAV and realize the accurate position adjustment of UAV,which verifies the effectiveness and feasibility of the positioning method proposed in this paper. |
PDF Full Text Request