Research on Real-Time Power Line Damage Detection Method Based on YOLO Algorithm
Tanyao Di, Liujing Feng, Huangyan Guo
Abstract
Because of the detection of power line damage status in rural and remote areas, a method based on the YOLO algorithm is put forward for power line damage detection. After preprocessing the pictures taken by UAV, YOLO is used to establish the target detection model for damaged power lines, optimize the data structure through data augmentation and data classification technology, and carry out training using multi-layer convolution operation and eigenvalue pyramid structure. The test results indicate that the model can detect different power line targets in various environments, and the complete accuracy of the experiment reaches 91.51%. The established UAV patrol mode and back office operation interface can be competent for conducting regular inspections of electric equipment in rural and remote areas.