Improved YOLOv8-based insulator defect detection system for transmission lines

Abstract During the long-term operation of high-voltage transmission lines, insulators are susceptible to defects such as flashover and structural breakage, which pose significant risks to the safety and stability of modern power grids. Traditional manual inspections and early computer vision-based approaches often struggle with low detection accuracy and high false positive rates, especially in complex environments and small-object scenarios. To overcome these shortcomings, this research presents a new insulator defect detection approach based on an improved YOLOv8 framework. The model embeds an Adaptive Feature Enhancement (AFE) module within the backbone network. This is aimed at enhancing the multi-scale contextual feature representation and strengthening the detection of subtle and small-sized defects. Additionally, the Convolutional Block Attention Module (CBAM) is employed to refine the model’s focus on defect-relevant regions while effectively suppressing background noise. The loss function is further optimized using a combination of Weighted IoU (WIoU) and a scale-adaptive weighting strategy, enhancing bounding box localization accuracy and overall robustness, particularly for small targets. Experimental results confirm the effectiveness of the proposed method. When compared with the baseline YOLOv8, it not only achieves a 2.2% improvement in mAP50 and a 1.4% enhancement in mAP50–95 but also significantly reduces the rates of missed detections and false alarms. By integrating advanced deep learning techniques into power system monitoring, this work demonstrates strong real-time performance and adaptability for intelligent inspection tasks. It contributes to the development of automated and intelligent maintenance frameworks within smart grid systems, particularly in the areas of artificial intelligence, intelligent automation, and control systems in electrical engineering. This research provides a robust and scalable solution for intelligent grid infrastructure monitoring.