Accurate and efficient insulator maintenance: A DETR algorithm for drone imagery

Yanfeng Tian, Rodina Binti Ahmad and Nor Aniza Binti Abdullah

PLOS ONE, 2025, vol. 20, issue 2, 1-18

Abstract: With the increasing demand for electricity, the safety and stability of power grids become paramount, highlighting the critical need for effective maintenance and inspection. Insulators, integral to power grid maintenance as protective devices on outdoor high-altitude conductors, are often subject to suboptimal image quality during drone-based inspections due to adverse weather conditions such as rain, snow, fog, and the challenges posed by sunlight, high-speed movement, and long-distance imaging. To address these challenges and achieve a more accurate inspection system, this manuscript introduces an insulator defect detection algorithm tailored for the low-quality images collected by drone-based imaging systems. Utilizing a patch diffusion model, high-quality images are obtained, enhancing the precision of insulator defect detection methods. Furthermore, to improve detection accuracy, we introduce an optimized DETR method that incorporates a Spatial Information Interaction Module to further strengthen the characteristics of minor defects. Additionally, a special convergence network is employed to augment the detection capabilities of the DETR. Experimental results demonstrate that our proposed insulator detection technique has achieved a detection accuracy of 95.8%, significantly outperforming existing defect detection methods in complex environments. It overcomes the drawbacks of traditional methods by employing sophisticated computational models, leading to more efficient, economical, and secure maintenance and inspection of power grids.

Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0318225 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 18225&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0318225

DOI: 10.1371/journal.pone.0318225

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().