A Study on the Grading Ring Recognition Method of Power Line Based on Deep Learning

It is an important measure to ensure the safe operation of the power system by using the machine vision with unmanned aerial vehicle to inspect the power lines. The grading ring detection is one of the tasks. For the captured images, there are complex issues such as complex and varied environmental backgrounds, inconsistent camera angles, difficulty in controlling the variation of light intensity, and jitter of drones, in which it is difficult to identify the grading ring. Based on the structure of convolutional neural network, a deep learning network structure is constructed for identifying and positioning of grading ring: the convolutional neural network method is used to extract the high-level characteristics of the grading ring; TensorFlow and Faster R-CNN are used to train the detection model of grading rings; Light intensity and motion blur experiments are designed, and the experiment results show that it is robust for the grading ring detection model. The high accuracy and good adaptability is verified in the actual test experiments, and the model can meet the needs of the grading ring identification and positioning in power lines in complex environments.