Research On Recognition System Of Lightning Arrester Monitor For Intelligent Operation And Inspection In Substation

To build a strong and reliable smart grid is the orientation of China’s power grid in 21~stt century.The inevitable trend of future power grid’s evolution includes the wide-area ubiquitous,open and joint sharing energy Internet which integrates multi-energy conversion technology,intelligent control technology and modern information technology.In order to reduce the missed-detection and false-detection that may be caused by the manual inspection of various instruments in the traditional substation,and in order to cooperate with the intellectualization of power grid,the recognition and reading system of the traditional analog instrument in the intelligent substation inspection is in urgent need of design.In this paper,a typical lightning arrester monitor is the object of the analog instrument image recognition system on the popular substation intelligent inspection robot.This paper,which strives to solve interference factors that affect machine recognition,including systematically researching the processing,correction and recognition system after the instrument image acquisition,is expected to be applied in engineering.For the hypothetical system above,this paper proposed a framework of instrument recognition of robot.The specific methods are as follows:1.Image preprocessing:Image gray processing is used to convert color image to gray space processing,image restoration method is used to correct image degradation;2.Image registration:extracting specific target from image with complex background,correcting partial image distortion;3.Camera pose estimation:estimate the attitude angle of the camera relative to the standard position,it’s used to correct severe image capture deviation and reading deviation;4.Pointer meter and digits reading recognition:digital image processing is used to identify the instrument readings.This paper is based on MATLAB R2018a.This paper studied the lightning arrester monitor recognition system based on computer vision and digital image processing in substation.Under the recognition framework proposed in this paper,the instrument image under standard viewing is obtained by preprocessing,image registration and reading error correction.Finally,the leakage current and the action times of the lightning arrester are identified by the pointer meter and the digits reading.The experimental results show that the effect of eliminating interference factors affecting instrument recognition is better.The experimental results show that the interference factors affecting the identification of the instrument are better.The 8 sets of data are tested and the maximum absolute error of the pointer reading does not exceed 0.021mA,which can meet the requirements of the national electrical instrument accuracy level 1.The digits are fully recognized.At the same time,the results of this paper show that the recognition system proposed in this paper is effective,and it can replace the manual inspection of the arrester instrument at a certain accuracy level.The system used on the inspection robot needs to be further optimized.At present,preliminary tests have been carried out on the inspection robot developed by the State Grid Sichuan Electric Power Research Institute in Zhongba 110kV substation.