A Insulator Defect Detection Method Based on YOLO and ChatGPT

Insulators are crucial components in power system transmission and distribution, operating in complex environments. Prolonged exposure to open-air conditions can lead to various faults, impacting the reliability of power supply and the overall safety and stability of the power system. This paper presents an insulator inspection system based on YOLOv8 and ChatGPT, which enhances the efficiency and accuracy of insulator defect detection. The system can automatically generate reports using ChatGPT, integrating traditional inspection methods with ChatGPT, thus aiding engineers in understanding and addressing faults. This paper proposes an improved YOLOv8 model that incorporates the LSKA attention mechanism and DCNv4 deformable convolution. An insulator detection system integrating ChatGPT and the enhanced YOLOv8 model is developed. Experimental results show that the improved model increases accuracy by 1.6% for dirty glass sheets, 2.2% for dirty polymer sheets, 5.6% for defective polymer sheets, and 3.1% for insulator flashover. The overall mAP50 improved by 1.2%. These results demonstrate that the improved YOLOv8 algorithm can accurately detect insulators, confirming the reasonableness and effectiveness of the enhancements.