Research on Detection Methods of Foreign Objects and Component Defects in Transmission Lines Based on Deep Learning

The foreign objects and component defects on the transmission line will cause adverse effects on the power transmission equipment, and may even endanger the safe operation of the power grid. Therefore, it is necessary to identify the foreign objects and component defects on the transmission line. In this paper, a method based on Faster-RCNN (Regional Convolutional Neural Network) for reorganization of the transmission line images collected by aerial photography is studied, where feature maps from the images to be detected through convolutional network operations is generated. Then the Region Proposal Network (RPN) is used to generate region of interests, and then the category of the picture is obtained. Moreover, the location of foreign objects and component defects through the classification layer is also recognized. The final result shows that the average accuracy of the presented method for the identification of foreign objects and component defects in transmission lines can reach 95.24%.