External Damage Identification Of Insulators Based On Convolutional Neural Network

Insulator is an important part of power network,which is responsible for supporting the circuit and forming good insulation between the current-carrying conductor and the ground.At the same time,the failure probability of insulator is higher,its uncleanness crack damage will affect the normal operation of the circuit,seriously will lead to tripping,blackout caused major losses,so for timely and accurate state detection of insulator is very be necessary at this stage detection method is mainly visual method,low efficiency and easily affected by subjective along with the development of the scale of transmission lines,eyeballing method cannot meet the needs of the electric power line inspection In this paper,image processing technology is applied to insulator state detection,and a method of insulator detection based on convolutional neural network is proposed.Specific research contents are as follows:(1)In order to improve the image quality,the median filter is used to reduce the image noise,then the histogram equalization is used to enhance the image contrast,and the image details are enhanced on the basis of effectively removing the image noise.(2)After preprocessing,the image segmentation algorithm is studied,different methods are simulated,and the image edge detection effect is evaluated according to the two indexes of edge continuity and edge order.Finally,the image is segmented by Laplacian operator,and the accurate image edge information is obtained.(3)This paper studies the deficiencies of the previous insulator identification methods,and through the learning of the operation steps and principles of the convolutional neural network,the convolutional neural network is applied to the insulator image classification,and the program is written by MATLAB.(4)MATLAB was used to carry out the experimental simulation of this method,and the accuracy and efficiency of classification were verified by using the field insulator images.The experimental data obtained proved that the accuracy of classification of this algorithm was better than the traditional algorithm.The accuracy and operation speed of this method for insulator identification were improved.The experimental simulation in this paper proves that the convolutional neural network can accurately judge whether the insulator is cracked or not,and the method is simple and effective,which provides a new idea and solution for the realization of smart power grid.