Real-time Condition Monitoring of Transmission Line Insulators Using the YOLO Object Detection Model with a UAV

• Published in: IEEE transactions on instrumentation and measurement Vol. 73; p. 1
• Main Authors: Panigrahy, Satyajit, Karmakar, Subrata
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.01.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Condition monitoring; Defects; Detectors; Feature extraction; Flashover; Image augmentation; Inspection; Insulators; Object recognition; Outdoor Insulators; Raspberry Pi 4 (IoT); Real time; Real-time systems; Remote Condition Monitoring; Single-stage Object Detection; Transmission lines; UAV (Hexacopter); Weather; YOLO; YOLO (You Only Look Once)
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2024.3381693

Continuous monitoring and inspection of high voltage insulators is necessary to prevent failures and emergencies. Manual inspections can be costly and time-consuming, particularly when covering large geographical areas exposed to harsh weather conditions. This study proposed a single-stage object detector approach to address the limitations of traditional inspection methods by utilizing a hexacopter for efficient inspections of outdoor insulators. The object detector model was trained using a dataset of 6020 insulator images for detecting defects in complex backgrounds. Image augmentation techniques were adopted to avoid overfitting. Finally, the hexacopter was equipped with an onboard camera and a Raspberry Pi 4 single-board computer to automate the outdoor insulator inspection system by detecting real-time defects. Experimental results demonstrated the effectiveness of the YOLOv8n object detector model in successfully identifying various insulator conditions, including normal, broken, polluted, and flashover surfaces, with a mAP@50 of 99.4%.