Research On Detection Method Of Transmission Line Insulator Based On Image Recognition

With the rapid deployment and development of Chinese electric power industry,grid security testing has been proposed continuously.As an important part of the power transmission line,the insulator device can provide a good insulation among the wire,the cross bar and the tower.However,with the long period of high-load operation and exposure to natural environments,insulators are prone to failures,such as cracks,icing,and stringing.Once a fault occurs,it will seriously hinder the stable operation of the power system.The traditional insulator inspection is a series of observations and records of the naked eye by the inspectors through the climbing tower.Due to the large national land span,the changing environmental conditions,the large layout of the transmission industry,and the large number of installation equipment,the workload of human detection is large,the cycle is long,and the safety of employees cannot be guaranteed,at the same time with all these problems it does not meet the development direction of the future smart grid.The rapid development of computer vision,using image acquisition equipment as a medium and image processing technology to identify and detect insulator devices,will greatly improve detection efficiency and reduce input costs.Therefore,this paper proposes an image-based power transmission equipment insulator detection method,which uses image recognition technology and classification algorithm instead of human detection,which not only improves work efficiency,but also protects workers' safety and meets the future development direction of smart grid industry.In view of the problem of insulator identification and positioning,above all,according to the disadvantages of ORB(Oriented FAST and Rotated BRIEF)algorithm,such as weak anti-viewing ability and high error rate,this paper proposes an improved ORB algorithm and experiments on the standard test set,so that the improved algorithm has better strong resistance to viewing angle change and better scale rotation invariance;for insulator images with multi-scale,multi-angle and rotation,etc.,constructing insulator identification and localization algorithm,first,the insulator image is preprocessed to obtain a binarized image,and then the morphological method is used to obtain a plurality of regions to be detected of the binarized image.Finally,the template insulator image is matched with the image to be detected by the improved ORB algorithm,and the region with the insulator string is identified according to the similarity measure.Through the experiment of the insulator image,the proposed algorithm,with high robustness,can identify and locate the insulator efficiently,and solve the problem of inaccurate recognition and location,due to factors such as large-angle and multi-scale,.Aiming at the problem of string loss of transmission line insulators,an algorithm based on Adaboost for insulator drop detection is used.Firstly of all,the insulator string with the rotation angle is adjusted to the vertical condition;then the insulators in the vertical state are respectively projected as the x-axis and the y-axis,with the width and height of the single insulator disk are obtained,thereby using a single disk height,width and the location information,a single disc region has been obtained by segmenting;finally,the feature information is extracted for each monolithic region,and the Adaboost classifier is used to classify whether the string is dropped,and the original image is marked according to the information of the dropped string position.The experimental result shows that the algorithm is accurate for the drop position mark and the detection accuracy is high.