Research method of insulator burst fault identification based on self-attention YOLOv4

The integrity of the insulator on the transmission line directly affects the safety and reliability of the transmission. In this paper, deep learning method is used to study the problem of insulator image recognition extraction and defect detection. Due to the limitation of the number of model parameters, the existing models cannot directly use high-resolution images, and the recognition accuracy has great room for improvement. In order to solve this problem, a new method of insulator burst fault identification based on self-attentional YOLOv4 is proposed by combining self-attentional YOLO with YOLO. Compared to other models of this type, Transformer’s self-attention mechanism can capture long distance dependencies in a sequence. Its model is able to take into account information that is far apart in the sequence to make more accurate predictions. Compared with the traditional YOLOv4 model, it can be found that the proposed insulator burst fault identification model can identify the insulation and its burst fault accurately and quickly.