A Method for Identifying Small-Size Defects of Insulators in Transmission Lines Based on the COK-YOLO Algorithm

Abstract

Insulators are an important guarantee for maintaining the stable operation of the power system. The faults caused by insulator defects will endanger the operation safety of the power system and may even lead to the breakdown of the system. Currently, there are various methods for detecting insulator damage. However, it is difficult for all the existing algorithms to detect the problem of small insulator defects. Aiming at the problem that small insulator defects are difficult to detect, this paper proposes a method for detecting small insulator defects based on the improved YOLOv8. This paper draws on the principle of Cross Stage Partial Network to put forward the COK-YOLO algorithm, and uses the Omni kernel to achieve multi-scale feature extraction, which improves the detection ability of small insulator defects and finally enhances the detection results of various types of defects including breakage and pollution flashover. The experimental results show that the improved algorithm strengthens the detection ability of small insulator defects and provides a guarantee for the safe operation of the power system.