Defect Detection of Solar Panels Based on UAV Perspective

DOI: https://doi.org/10.62517/jes.202502410

Author(s)

Tao Chen, Tianxiang Hou, Yang Cao, Wei Liu*

Affiliation(s)

KeWen College, Jiangsu Normal University, Xuzhou, Jiangsu, China *Corresponding Author

Abstract

The global energy structure is rapidly transitioning towards clean energy, and the solar power generation industry is developing quickly. If solar panels have defects, their power generation efficiency decreases, maintenance costs increase, and safety incidents may even occur. Therefore, this study proposes a scheme for detecting defects in solar panels by integrating drone technology, deep learning algorithms, and multi-source data processing. Using the drone's flight control system and optimised path planning system, images of large-scale panels are efficiently collected. With the improved YOLOv8 algorithm and multispectral image processing technology, even tiny defects such as hairline cracks can be accurately identified. Combined with 5G real-time transmission and distributed cloud storage, a comprehensive data management system is established. Experiments show that under different lighting conditions and terrains, the system achieves a defect detection accuracy of 92.3% and a recall rate of 87.6%. Its efficiency is more than 15 times higher than manual inspection and can reduce maintenance costs by 30%, thereby providing the industry with an intelligent and cost-effective solution.

Keywords

Drone; Solar Panels; Defect Detection; Deep Learning; YOLOv8; Multispectral Imaging

References

[1]Li Yunxiang. Research on Fault Detection Method of UAV Inspection for Photovoltaic Panels Based on Deep Learning. Xi'an Shiyou University, 2025. DOI:10.27400/d.cnki.gxasc.2025.000504. [2]Hou Jinlong, Xu Bin, Liu Shaoguang. Application and Analysis of UAV-Mounted High-Definition Visible Light and Infrared Thermal Imaging for Exterior Wall Leakage Detection. China Building Waterproofing, 2024, (06):39-42. DOI:10.15901/j.cnki.1007-497x. 2024. 06. 010. [3]Zhang Shaohui, Shi Yanxiong. Application of 5G Edge Computing in Real-Time Data Processing. China Broadband, 2025, 21(08):106-108. DOI: 10.20167/ j. cnki. ISSN1673-7911.2025.08.36. [4]Wang Duanfa. Enhancing CLAHE using interval-valued Fermatean fuzzy sets for robust low-light image enhancement. Glass, 2025, 52(07):13-20. [5]Bao Yidong, Li Lin, Su Xiaowei. Improved median filtering and adaptive threshold canny edge detection algorithm. Science and Innovation, 2025, (07):31-33. DOI: 10.15913/j.cnki.kjycx.2025.07.008. [6]Yang Wei, Zhang Changsheng, Liu Hui. YOLOv8-DM Lightweight Photovoltaic Module Defect Detection Method. Journal of National University of Defense Technology, 2025, 47(04):158-169. [7]An Junxiu, Jin Yuchang, Yang Linwang, et al. Fundamentals and Practice of Hadoop Big Data Processing Technology [M]. People's Posts and Telecommunications Press: 2024 08:254. [8]Xu Mengying. Research on Key Technologies of Self-Organising Network for Firefighting Drones Based on Swarm Intelligence Algorithms. Beijing University of Posts and Telecommunications, 2025. DOI:10.26969/d.cnki.gbydu.2025.000018. [9]Huang Yi, Ma Yili, Miao Ankang, et al. Combined Photovoltaic Power Output Prediction Model Based on GSS RIME-CNN-LSTM-Attention. Acta Energiae Solaris Sinica, 2025, 46(10): 250-258. DOI: 10.19912/j.0254-0096.tynxb.2024-1036. [10]Li Xianglong. Research on Key Technologies of Parallel Learning for Multi-UAV Cooperative Operations. Tianjin University, 2020. DOI: 10.27356/ d.cnki.gtjdu.2020.004552.