Design Of UAV Power Inspection System Based On Insulator Self-explosion Detection

With the continuous progress of science and technology and the rapid development of modern power system,robot has gradually replaced manual inspection in the power inspection task of substation,which is also the inevitable trend of the development of modern power system to smart grid.However,most of the current inspection work is done manually,which has the defects of low degree of information and poor inspection effect.Therefore,the design and development of a complete UAV power inspection system can overcome the existing problems,reduce the labor intensity of inspectors,and improve the efficiency and accuracy of inspection results,which has important engineering application value.There are many inspection items in power line inspection.Insulator string is the key device in high voltage transmission line.Because it is exposed outdoors for a long time,it is prone to failure,and it is the monitoring object in power line inspection.Based on the literature review of power inspection technology at home and abroad,combined with the specific application of current power inspection work,this paper analyzes the business requirements,user requirements and functional requirements of power inspection.The function design and implementation of UAV power inspection system including power inspection management terminal,power inspection terminal and insulator self explosion detection are completed.Based on B / S architecture,seven modules are developed,including user login module,user management module,task management module,map module,patrol data management module,patrol result recognition module and fault alarm module.Web pages are provided for patrol task manager to manage tasks and query results.Based on C / S architecture,six modules are developed,which are terminal user login module,task query module,result upload module,map positioning module,communication module and weather query module.App interface is provided for inspection staff to operate.A "two-stage multi model" fault detection algorithm for insulator self explosion is proposed.The algorithm divides the problem of insulator self explosion fault detection into two stages: insulator string bead identification and self explosion defect identification.The cyclegan model is used for style migration data augmentation,and the mask model is used The r-cnn model is used to segment the image semantic of the insulator string bead area.Combined with the image processing technology,the connected area of the insulator string bead is obtained.Finally,the yolov4 model is used to detect the self explosion defect of the insulator to complete the detection of the self explosion fault of the insulator.In this paper,the UAV power inspection system realizes a series of operations from the management side to make inspection tasks,to the terminal to receive tasks,to complete tasks,to send inspection results photos to the management side,to automatically identify the insulator self explosion area,and to query inspection results.The overall debugging of the system is realized,and the feasibility of the system is verified.