Deep Learning-Based Fault Detection in Electrical Insulators Using Object Detection Models

This study presents a deep learning-based approach for detecting faults in electrical insulators used in high-voltage power transmission lines. The proposed system utilizes the YOLOv8 object detection algorithm and is trained on a dataset of 239 labeled images, classified as either “intact” or “faulty.” The dataset comprises images captured under diverse environmental conditions, including variations in illumination and viewing angles. To ensure a robust evaluation, the performances of YOLOv8, YOLOv12, and YOLO-NAS models were compared. Preprocessing involved data augmentation techniques such as rotation, flipping, and brightness/contrast adjustments. All models were trained using the same hyperparameters and evaluated with precision, recall, F1-score, and mean average precision (mAP) metrics. The results demonstrate that YOLOv8 outperforms the other models in both detection accuracy and real-time inference capability. Overall, this study highlights the potential of modern object detection architectures for automating fault diagnosis in power infrastructure, significantly improving inspection speed and reliability.