Journals / Harran Üniversitesi Mühendislik Dergisi / 2026 / Cilt: 11 Sayı: 1

A Deep Learning-Based Approach for Fault Detection in Solar Panels: Application of YOLOv8

Güneş Panellerindeki Arıza Tespiti için Derin Öğrenmeye Dayalı Bir Yaklaşım: YOLOv8 Uygulaması

Husseın Yousefalturk(DİCLE ÜNİVERSİTESİ, MÜHENDİSLİK FAKÜLTESİ)0000-0002-5237-4448
Yurdagül Benteşen Yakut(DİCLE ÜNİVERSİTESİ, MÜHENDİSLİK FAKÜLTESİ)0000-0003-3236-213X
Pages
54–68
Publication Date
2026-03-31
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Experimental results indicate that the proposed approach achieves a 94.3% [email protected], along with 89% Top-1 accuracy and 99.1% Top-5 accuracy.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Factors such as bird droppings, dust accumulation, physical/electrical faults, and snow coverage significantly reduce the energy production efficiency of photovoltaic panels.
📊
Key Findings & Data:This study proposes an integrated YOLOv8-based deep learning framework that enables high-accuracy detection and classification of multiple fault types within a single processing pipeline and with low computational cost.
💡
Conclusion & Impact:Experimental results indicate that the proposed approach achieves a 94.3% [email protected], along with 89% Top-1 accuracy and 99.1% Top-5 accuracy.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Derin öğrenme · YOLOv8 · Güneş paneli

Abstract

Factors such as bird droppings, dust accumulation, physical/electrical faults, and snow coverage significantly reduce the energy production efficiency of photovoltaic panels. This study proposes an integrated YOLOv8-based deep learning framework that enables high-accuracy detection and classification of multiple fault types within a single processing pipeline and with low computational cost. A total of 887 high-resolution images obtained from the Kaggle platform were annotated using Roboflow, and data augmentation techniques including rotation, flipping, and brightness adjustment were applied to enhance dataset diversity and generalization. The model was trained in the Google Colab environment using pre-trained weights with a 70% training, 20% validation, and 10% test split. Experimental results indicate that the proposed approach achieves a 94.3% [email protected], along with 89% Top-1 accuracy and 99.1% Top-5 accuracy. The key contribution of this study lies in leveraging the unified architecture of YOLOv8 to jointly optimize object detection and classification tasks without employing separate models. The findings demonstrate that the proposed framework can be effectively integrated into real-time monitoring and autonomous maintenance systems for large-scale solar power plants, offering a scalable, reliable, and application-oriented solution that contributes to the existing literature.

Özet

Güneş panellerinde oluşan kuş pisliği, tozlanma, fiziksel/elektriksel hasarlar ve kar örtüsü gibi faktörler, enerji üretim verimliliğini ciddi biçimde düşürmektedir. Bu çalışmada, söz konusu arıza türlerinin tek bir işlem hattı içerisinde, düşük hesaplama maliyetiyle ve yüksek doğrulukla tespit ve sınıflandırılmasını hedefleyen, YOLOv8 tabanlı bütünleşik bir derin öğrenme yaklaşımı önerilmektedir. Kaggle platformundan temin edilen 887 adet yüksek çözünürlüklü görüntü, Roboflow ortamında etiketlenmiş; veri setinin çeşitliliğini ve genelleme yeteneğini artırmak amacıyla döndürme, çevirme ve parlaklık ayarı gibi veri artırma teknikleri uygulanmıştır. Model eğitimi, %70 eğitim, %20 doğrulama ve %10 test veri bölünmesiyle, önceden eğitilmiş ağırlıklar kullanılarak Google Colab ortamında gerçekleştirilmiştir. Deneysel sonuçlar, önerilen modelin %94,3 [email protected], %89 Top-1 doğruluk ve %99,1 Top-5 doğruluk değerlerine ulaştığını göstermektedir. Çalışmanın özgün katkısı, YOLOv8’in birleşik mimarisinden yararlanarak nesne tespiti ve sınıflandırma görevlerini ayrı modeller kullanmaksızın tek bir yapı altında optimize etmesidir. Elde edilen bulgular, geliştirilen yaklaşımın büyük ölçekli güneş enerjisi santrallerinde (GES) gerçek zamanlı izleme ve otonom bakım sistemlerine entegre edilebilir, ölçeklenebilir ve güvenilir bir çözüm sunduğunu ortaya koymaktadır. Bu yönüyle çalışma, saha uygulamalarına yönelik pratik bir çerçeve sunarak literatüre katkı sağlamaktadır.

🔖

Cite This Article

Yousefalturk, H., Yakut, Y. B. (2026). A Deep Learning-Based Approach for Fault Detection in Solar Panels: Application of YOLOv8. Harran Üniversitesi Mühendislik Dergisi, Cilt: 11 Sayı: 1, 54–68. https://doi.org/10.46578/humder.1821648

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (29)

  1. REN21, Renewables 2023 Global Status Report, Renewable Energy Policy Network for the 21st Century, 2023.
  2. Kannan, R., Leong, K. C., Osman, R., Ho, H. K., & Tso, C. P. (2006). Life cycle assessment study of solar PV systems: An example of a 2.7 kWp distributed solar PV system in Singapore. Solar energy, 80(5), 555-563.
  3. Mekhilef, S., Saidur, R., & Kamalisarvestani, M. (2012). Effect of dust, humidity and air velocity on efficiency of photovoltaic cells. Renewable and sustainable energy reviews, 16(5), 2920-2925.
  4. Li, B., Delpha, C., Diallo, D., & Migan Dubois, A. (2021). Application of artificial neural networks to photovoltaic fault detection and diagnosis: A review. Renewable & Sustainable Energy Reviews, 138, 110512. https://doi.org/10.1016/j.rser.2020.110512
  5. LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436-444.
  6. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 25.
  7. Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 779-788).
  8. Ultralytics, YOLOv8: A New State-of-the-Art in Object Detection, [Online] (2023). Erişim: https://docs.ultralytics.com.
  9. Önder, Ö., & Karan, Y. (2024). Çay ve Eğrelti Otunun YOLOv5 ve YOLOv8 Algoritmaları ile Karşılaştırmalı Tespiti. Recep Tayyip Erdogan University Journal of Science and Engineering, 5(1), 74-88.
  10. Alemdar, K. D. (2024). Çift Sıra Parklanma Durumunun Nesne Tespit Algoritması YOLOv8 ile Tespit Edilmesi. Journal of the Institute of Science and Technology, 14(3), 1164-1176.
  11. Budak, İ., Bal, S., & Korkmaz, H. (2025). PCB Üretiminde Çok Sınıflı Kusur Tespiti için YOLO Tabanlı Derin Öğrenme Modeli. Afyon Kocatepe Üniversitesi Fen Ve Mühendislik Bilimleri Dergisi, 25(4), 816-826.
  12. Gu, K., & Chen, Y. (2024). YOLOv3-MSSA based hot spot defect detection for photovoltaic power stations. Journal of Measurements in Engineering, 12(1), 23-39.
  13. Di Tommaso, A., Betti, A., Fontanelli, G., & Michelozzi, B. (2022). A multi-stage model based on YOLOv3 for defect detection in PV panels based on IR and visible imaging by unmanned aerial vehicle. Renewable energy, 193, 941-962.
  14. Zhang, M., & Yin, L. (2022). Solar cell surface defect detection based on improved YOLO v5. IEEE access, 10, 80804-80815.
  15. Lei, Y., Wang, X., An, A., & Guan, H. (2024). Deeplab-YOLO: A method for detecting hot-spot defects in infrared image PV panels by combining segmentation and detection. Journal of Real-Time Image Processing, 21(2), 52.
  16. Batool, A., Kim, Y. W., & Byun, Y. C. (2025). Improved YOLOv8 framework for efficient solar panel defect detection. Journal of Building Engineering, 111, 113031.
  17. Vaghela, R., Vaishnani, D., Srinivasu, P. N., Popat, Y., Sarda, J., Woźniak, M., & Ijaz, M. F. (2025). Land cover classification for identifying the agriculture fields using versions of yolo v8. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.
  18. Zhang, L., Wu, X., Liu, Z., Yu, P., & Yang, M. (2024). ESD-Yolov8: an efficient solar cell fault detection model based on Yolov8. IEEE Access, 12, 138801-138815.
  19. Ye, K., & Xue, Y. (2023, November). Image recognition of garbage classification based on YOLOv8. In 7th International Conference on Vision, Image and Signal Processing (ICVISP 2023) (Vol. 2023, pp. 145-149). IET.
  20. Haeruman, A., Haq, S. U., Mohandes, M., Rehman, S., & Mıtu, S. S. I. (2024). Ai-based pv panels inspection using an advanced yolo algorithm. Materials Research Proceedings, 43.
  21. Kaggle, Kaggle Datasets, [Çevrimiçi]. (2025). Erişim: https://www.kaggle.com/datasets.
  22. Roboflow, Roboflow Universe, [Çevrimiçi].(2025) Erişim: https://universe.roboflow.com.
  23. G. Jocher, A. Chaurasia, J. Qiu, Ultralytics YOLO, [Online]. (2023). Erişim: https://github.com/ultralytics/ultralytics.
  24. Li, Y., Fan, Q., Huang, H., Han, Z., & Gu, Q. (2023). A modified YOLOv8 detection network for UAV aerial image recognition. Drones, 7(5), 304.
  25. Reis, D., Kupec, J., Hong, J., & Daoudi, A. (2023). Real-time flying object detection with YOLOv8. arXiv preprint arXiv:2305.09972.
  26. Fawcett, T. (2006). An introduction to ROC analysis. Pattern recognition letters, 27(8), 861-874.
  27. Powers, D. M. (2020). Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. arXiv preprint arXiv:2010.16061.
  28. Everingham, M., Van Gool, L., Williams, C. K., Winn, J., & Zisserman, A. (2010). The pascal visual object classes (voc) challenge. International journal of computer vision, 88(2), 303-338.
  29. Kılıç, K., Özcan, U., Kılıç, K., & Dogru, İ. (2024). Using deep learning techniques furniture image classification. Politeknik Dergisi, 27(5), 1903-1911.