Transmission line insulator defect detection based on computer vision

  • Dr Minh Nguyen Quoc School of Electrical and Electronics Engineering, Hanoi University of Science and Technology
  • Nguyễn Thị Quỳnh Faculty of Information Technology, Hanoi University of Industry
Keywords: Insulator; Transmission line; Computer vision; Faster R-CNN; YOLOv8.

Abstract

The quick and accurate detection of faults in insulator strings plays a crucial role in ensuring the stability and reliability of the transmission power grid. In this study, the authors propose a method for detecting and classifying insulation faults on transmission line insulator strings based on computer vision models. The dataset used in the study consists of 1,600 images of insulator strings with a total of 18,195 objects categorized into three types: normal functioning strings, cracked strings, and strings damaged by lightning strikes. The dataset is split into a training set and a test set with an 80-20 ratio. The dataset is trained on two computer vision models, YOLOv8 and Faster R-CNN. The results show that the YOLOv8 model can detect and classify cracked and lightning-damaged insulator strings with an accuracy of up to 96.4%.

điểm /   đánh giá
Published
2023-10-25
Issue
Vol. 90 (2023)
Section
Overview