Defect Recognition And Diagnosis Of Composite Insulator Shed Based On Visible Light Image Processing

Composite insulators are widely used in power systems due to their small size,low cost,high strength,and pollution flashover resistance.However,as the service life increases,composite insulators will produce aging phenomena such as crazing of the shed,the chalking of the insulator and the decrease of hydrophobicity.If these aging problems are not dealt with in time,it will lead to accidents such as flashover,brittle fracture or even string breakage of insulators,which will seriously affect the power transmission safety of the power grid.At present,the identification and diagnosis of crazing defects and chalking defects requires a lot of manpower and material resources.Therefore,this paper analyzed the visible light images of composite insulator shed that have been in operation for many years from the perspective of image processing,and studies the shed crazing and chalking defects of composite insulators.For the crazing phenomenon of shed of composite insulator,first,the insulator shed image that has been running for a long time in the field has been shot in the laboratory.Secondly,the insulator shed image is preprocessed,including image grayscale,binarization and connected area marking,which automatically segment the shed image.A variety of edge extraction operators are compared,and the operator with better effects is selected for feature of the crazing defect.Then the crazing phenomenon was characterized by the fractal dimension,and the result showed that the higher the calculated value of the fractal dimension,the more serious the shed crazing phenomenon,and the diagnosis of the shed crazing defect was realized.For the chalking defects of composite insulator sheds,firstly,the shed images of the new insulators and the sheds with chalking defects after long-term operation were taken in the laboratory.In order to extract the insulator region,this paper preprocesses the image,including image color space decomposition and image multiplication.After that,different color space components are extracted,and the color space components are screened by linear discriminant analysis method,and the characteristic quantities that can characterize chalking defects are obtained.The support vector machine algorithm was used to identify the chalking defects.The experimental results show that the recognition rate of chalking defects is 97%.