Research On Key Technologies Of Image Processing For Substation Patrol Robot

In today's power system,substation,as the transfer station of power transmission and distribution,plays a key role in transforming voltage,transferring electric energy and distributing energy,and is the bridge between power station and users of power grid.In substations,power equipment is the key equipment of transmission and distribution.The stability and safety of equipment is related to the daily operation of the substation.The typical appearance defects of equipment and the recording of instrument data have been developed to automation and intellectualization,and the inspection task has been completed by combining the image processing algorithm.Therefore,it is of great significance to study the key technologies involved in substation inspection robot.This paper focuses on the image processing technology of substation inspection robot.First,the substation inspection robot system is designed to have high-speed processors and various sensors,which can identify and analyze various physical information including instrument data and equipment status.The robot is connected with the remote client through the Internet.With the remote client,the robot can be directly operated by supervisors and display the image,video,system status,sensor data.In the hardware module,the robot includes three parts: motor drive,ultrasonic ranging and platform control.For accuracy,the robot uses the PID algorithm as the mobile control algorithm,and the high precision gyroscope is used to measure the horizontal attitude of the robot when it moves.Secondly,the paper analyses the influence of external factors such as positioning error,mechanical wear and illumination environment on the recognition of typical appearance defects and instrument pointer.Image matching method is studied to eliminate the visual offset caused by positioning error and mechanical wear,including template matching,SHIT transformation,SURF transformation,and the effect is verified by experiments.On this basis,the paper obtains surface defect recognition through background noise modeling,and designs a Hough transform pointer recognition method based on region partitioning.Finally,the inspection robot system for substation is tested,and the visual terminal control system runs normally.At the same time,experiments are carried out on the typical appearance defect recognition method based on background noise modeling and the block Hough transform detection method,and the results show that the recognition accuracy can be effectively improved.The research of this paper enriches the image processing technology of substation inspection robot,has certain theoretical significance and engineering application reference value.