Solar Panel Defect Detection and Panel Localization Using Yolov5

DOI

10.2991/978-94-6463-589-8_15How to use a DOI?

Keywords

Solar Panel; Defect Detection; YOLOv5; localization

Abstract

Large-scale solar farms, often encompassing more than 100,000 panels, face significant challenges in maintaining optimal performance due to the impracticality of manual defect inspections. Undetected defects reduce panel efficiency, increase operational costs, and delay necessary maintenance, leading to more extensive damage and higher repair costs. This paper outlines a comprehensive approach to automatically detect defects and localize both normal and defective solar panels using the YOLOv5 model, addressing the need for efficient and reliable maintenance in large-scale solar farms. Initially, YOLOv5 is employed to classify specific zones within images containing two panels. Identified zones are cropped, and the same YOLOv5 model is used again to accurately localize each individual panel within the zone. Subsequently, the model is reapplied to detect any defects in the solar panels, analyzing and identifying anomalies. The panels and their defects are then precisely located, with bounding boxes drawn around the defect spots. The proposed method ensures thorough and precise identification and localization of both the panels and their defects. The final training results demonstrate near-perfect performance across all metrics, achieving a precision (P) of 0.947, a recall (R) of 0.968, and a mean Average Precision at 50% IoU (mAP50) of 0.989 for all classes. This project addresses critical challenges in the maintenance of large-scale solar farms, enhancing the efficiency and longevity of solar panels through timely and accurate defect detection. The automated system reduces labor costs, minimizes downtime, and promotes sustainable energy production. By fostering innovation in AI and image processing, the project contributes to technological advancements and supports global transitions to renewable energy sources. Future efforts will focus on real-time deployment on edge devices, integration with maintenance systems, expanding datasets for improved model robustness, and exploring multispectral imaging. Additionally, efforts will be made to integrate predictive maintenance algorithms, and conduct extensive field testing and long-term validation.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Muhammad Yunus Iqbal Basheer
AU  - Azliza Mohd Ali
AU  - Nurzeatul Hamimah Abdul Hamid
AU  - Mohd Zaliman Mohd Yusoff
PY  - 2024
DA  - 2024/12/01
TI  - Solar Panel Defect Detection and Panel Localization Using Yolov5
BT  - Proceedings of the International Conference on Innovation & Entrepreneurship in Computing, Engineering & Science Education (InvENT 2024)
PB  - Atlantis Press
SP  - 139
EP  - 149
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-589-8_15
DO  - 10.2991/978-94-6463-589-8_15
ID  - Basheer2024
ER  -