Research Article
Design of Porcelain Insulator Defect Recognition System Based on UAV Line Inspection Image
@INPROCEEDINGS{10.1007/978-3-031-04245-4_35, author={Zhaoyu Li and Zhong Zheng and Shuo Shi and E. Rui}, title={Design of Porcelain Insulator Defect Recognition System Based on UAV Line Inspection Image}, proceedings={6GN for Future Wireless Networks. 4th EAI International Conference, 6GN 2021, Huizhou, China, October 30--31, 2021, Proceedings}, proceedings_a={6GN}, year={2022}, month={5}, keywords={UAV Image processing Semantic segmentation}, doi={10.1007/978-3-031-04245-4_35} }- Zhaoyu Li
Zhong Zheng
Shuo Shi
E. Rui
Year: 2022
Design of Porcelain Insulator Defect Recognition System Based on UAV Line Inspection Image
6GN
Springer
DOI: 10.1007/978-3-031-04245-4_35
Abstract
With the progress of technology and the improvement of equipment quality, the coverage of China’s transmission network is expanding rapidly. Large power grids across complex and volatile terrain and dangerous high-voltage transmission lines are also being extended. Therefore, the traditional method of checking the circuit manually is no longer feasible due to its low efficiency, low precision, high risk and high cost. However, unmanned aerial vehicles (uavs) are a perfect way to circumvent these problems by inspecting transmission lines instead of workers. This paper takes the application of unmanned aerial vehicle in power line patrol as the research background, takes the porcelain vase in power transmission as an example, and realizes the image recognition and damage judgment system of the porcelain vase with specific target. Based on the image processing technology of machine learning and MATLAB, the target detection method of YOLO v3, the semantic segmentation method of Deeplab v3+, and the improved damage analysis method of ellipse fitting were respectively used to make the damage judgment and analysis of porcelain vats based on the intelligent image recognition interception, contour extraction and semantic segmentation. In the actual site of 166 porcelain bottles damage detection, damage detection accuracy reached 86.7%. Finally, the identification system of porcelain vase defect is realized.
