A Hybrid Deep Feature Fusion and CWINCA-Based Classification Framework for Thermal Fault Diagnosis in Photovoltaic Panels

Öz

Accurate and timely identification of faults in photovoltaic (PV) panels is critical for maintaining system efficiency and ensuring safe operation. In this study, a hybrid classification framework is proposed that integrates deep feature fusion with an advanced feature selection method to detect PV panel faults using thermal infrared imagery. Feature representations were extracted using four pre-trained lightweight convolutional neural networks: MobileNet, MobileNetV2, MobileNetV3Small, and MobileNetV3Large resulting in a 3840-dimensional concatenated feature vector. To reduce redundancy and improve discriminative power, the Cumulative Weight-based Iterative Neighborhood Component Analysis (CWINCA) was employed, selecting 142 informative features. These were subsequently classified using a linear Support Vector Machine (SVM). Experiments were conducted on the publicly available PVF-10 dataset, comprising 5,579 thermal images across ten fault categories. The proposed method achieved an overall classification accuracy of 86.49%, outperforming several individual CNN based architectures. The results demonstrate that combining feature-level integration with targeted selection significantly enhances classification performance while maintaining low computational complexity. This framework offers a promising and scalable solution for UAV-based PV inspection systems.

Anahtar Kelimeler

• Photovoltaic fault classification
• Thermal infrared imagery
• MobileNet
• Feature fusion
• CWINCA; Support vector machine.

Etik Beyan

“There is no need for an ethics committee approval in the prepared article” “There is no conflict of interest with any person/institution in the prepared article”

Termal Görüntülerle Fotovoltaik Panel Arızalarının Sınıflandırılmasına Yönelik CWINCA Tabanlı Hibrit Derin Öznitelik Birleştirme Yaklaşımı

Öz

Fotovoltaik panellerde meydana gelen arızaların doğru ve zamanında tespiti, sistem verimliliğinin korunması ve güvenli çalışmanın sürdürülebilirliği açısından kritik öneme sahiptir. Bu çalışmada, termal kızılötesi görüntülerden PV panel arızalarını sınıflandırmak amacıyla derin öznitelik birleştirme ve gelişmiş bir öznitelik seçimi yöntemini entegre eden hibrit bir sınıflandırma çerçevesi önerilmiştir. Öznitelikler, önceden eğitilmiş dört hafif konvolüsyonel sinir ağı (MobileNet, MobileNetV2, MobileNetV3Small ve MobileNetV3Large) kullanılarak çıkarılmış ve 3840 boyutunda birleşik bir öznitelik vektörü oluşturulmuştur. Ardından, öznitelik uzayındaki fazlalıkları azaltmak ve ayırt edici gücü artırmak amacıyla Kümülatif Ağırlık Tabanlı Yinelemeli Komşuluk Bileşen Analizi (CWINCA) yöntemi uygulanmış ve 142 anlamlı öznitelik seçilmiştir. Bu öznitelikler, doğrusal destek vektör makinesi sınıflandırıcısı ile sınıflandırılmıştır. Yöntem, kamuya açık PVF-10 veri kümesi üzerinde değerlendirilmiş ve on farklı arıza türünü içeren 5.579 termal görüntü ile test edilmiştir. Önerilen yöntem, %86.49 genel doğruluk oranı elde etmiş ve birçok bireysel CNN tabanlı mimarinin üzerinde performans göstermiştir. Sonuçlar, öznitelik düzeyinde birleştirme ile hedeflenmiş seçim yöntemlerinin bir arada kullanılmasının sınıflandırma başarımını artırırken hesaplama karmaşıklığını düşük tuttuğunu göstermektedir. Bu çerçeve, insansız hava araçları ile gerçekleştirilen PV denetim sistemleri için ölçeklenebilir ve etkili bir çözüm sunmaktadır.

Anahtar Kelimeler

• Fotovoltaik arıza sınıflandırması
• Termal kızılötesi görüntüleme
• MobileNet
• Öznitelik birleştirme
• CWINCA; Destek vektör makineleri.

Etik Beyan

"Hazırlanan makalede etik kurul onayına gerek yoktur." "Hazırlanan makalede herhangi bir kişi/kurumla çıkar çatışması bulunmamaktadır."

Kaynakça

1. G. Masson, A. Jäger-Waldau, I. Kaizuka, J. Lindahl, J. Donoso, and M. de l'Epine, "A snapshot of the global PV market," in 2024 IEEE 52nd Photovoltaic Specialist Conference (PVSC), 2024, pp. 566–568.
2. A. Jäger-Waldau, "Snapshot of photovoltaics March 2025," EPJ Photovoltaics, vol. 16, p. 22, 2025.
3. H. S. Muttashar, and A. M. Shakir, "Enhancing PV fault detection using machine learning: Insights from a simulated PV system," 2024.
4. A. Thakfan, and Y. Bin Salamah, "Artificial-intelligence-based detection of defects and faults in photovoltaic systems: A survey," Ener., vol. 17, no. 19, p. 4807, 2024.
5. C. M. Bohra, R. M. Srivastava, M. Aeri, and S. A. Dhoundiyal, "Identification of solar faults using machine learning," in 2024 International Conference on Cybernation and Computation (CYBERCOM)*, 2024, pp. 262–268.
6. M. Abdelsattar, A. AbdelMoety, and A. Emad-Eldeen, "Comparative analysis of machine learning techniques for fault detection in solar panel systems," SVU-Inter. Jour. of Eng. Scie. and Appl., vol. 5, no. 2, pp. 140–152, 2024.
7. Y. Ledmaoui, A. El Maghraoui, M. El Aroussi, and R. Saadane, "Enhanced fault detection in photovoltaic panels using CNN-based classification with PyQt5 implementation," Sens., vol. 24, no. 22, p. 7407, 2024.
8. M. I. M. Ameerdin, M. H. Jamaluddin, A. Z. Shukor, and S. Mohamad, "A review of deep learning-based defect detection and panel localization for photovoltaic panel surveillance system," Inter. Jour.of Rob.s and Cont.Syst., vol. 4, no. 4, pp. 1746–1771, 2024.

9. S. R. Joshua, S. Park, and K. Kwon, "EfficientNetB0 for AI-powered solar panel maintenance: Advanced fault detection," in 2024 15th International Conference on Information and Communication Technology Convergence (ICTC), 2024, pp. 301–306.
10. N. Elyanboiy, O. Eloutassi, M. Khala, I. Elabbassi, N. Elhajrat, and Y. Elhassouani, "Advanced intelligent fault detection for solar panels: Incorporation of dust coverage ratio calculation," in 2024 4th International Conference on Innovative Research in Applied Science, Engineering and Technology (IRASET), 2024, pp. 1–6.
11. E. Lodhi, X. Liu, G. Xiong, A. Khan, Z. Lodhi, T. Nawaz, A. Dilawarr, and F.-Y. Wang, "An efficient AIoT-based framework for remote monitoring and fault diagnostics in photovoltaic systems," SSRN, [Online]. Available: https://ssrn.com/abstract=5079421.
12. B. Anish, B. Gokul, M. Harish, and P. Kavin, "Solar panel fault detection using Internet of Things," in 2024 International Conference on IoT Based Control Networks and Intelligent Systems (ICICNIS), 2024, pp. 727–731.
13. N. Le, H. Vu, N. Porntipworawech, S. Waisayarat, and M. Doan, "Integration of aerial thermal imaging and deep learning for fault detection in photovoltaic panels: A study at Thinh Long solar power plant," in 2024 International Conference on Smart Energy Systems and Technologies (SEST), 2024, pp. 1–6.
14. W. Tang, Q. Yang, Z. Dai, and W. Yan, "Module defect detection and diagnosis for intelligent maintenance of solar photovoltaic plants: Techniques, systems and perspectives," Ener., p. 131222, 2024.
15. I. Polymeropoulos, S. Bezyrgiannidis, E. Vrochidou, and G. A. Papakostas, "Enhancing solar plant efficiency: A review of vision-based monitoring and fault detection techniques," Techn., vol. 12, no. 10, p. 175, 2024.
16. S. Govindarajan and V. Subramanian, "Enhanced fault detection and classification in solar PV systems using RNN and IoT integration," in 2024 10th International Conference on Electrical Energy Systems (ICEES), 2024, pp. 1–4.
17. M. Islam, M. R. Rashel, M. T. Ahmed, A. K. Islam, and M. Tlemçani, "Artificial intelligence in photovoltaic fault identification and diagnosis: A systematic review," Ener., vol. 16, no. 21, p. 7417, 2023.
18. D. Marangis, A. Livera, G. Tziolis, G. Makrides, A.Kyprianou, and G.E. Georghiou, "Trend-based predictive maintenance and fault detection analytics for photovoltaic power plants," Solar RRL, vol. 8, no. 24, p. 2400473, 2024.
19. M. Bouzidi, M. B. Rahmoune, and A. Nasri, "Intelligent fault detection of photovoltaic panel using neural networks," Stud.in Eng. and Exact Sci., vol. 5, no. 1, pp. 3161–3177, 2024.
20. A.G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam, "Mobilenets Efficient convolutional neural networks for mobile vision applications," arXiv preprint arXiv:1704.04861, 2017.
21. M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, "Mobilenetv2: Inverted residuals and linear bottlenecks," in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2018, pp. 4510–4520.
22. A. Howard, M. Sandler, G. Chu, L.-C. Chen, B. Chen, M. Tan, W. Wang, and Y. Zhu, Ruoming Pang, Vijay Vasudevan, Quoc V. Le, Hartwig Adam"Searching for mobilenetv3," in Proc. IEEE/CVF Int. Conf. Computer Vision (ICCV), 2019, pp. 1314–1324.
23. N. V. Sridharan, and V. Sugumaran, "Visual fault detection in photovoltaic modules using decision tree algorithms with deep learning features," Ener. Sour., Part A: Recovery, Utilization, and Environmental Effects, vol. 47, no. 2, p. 2020379, 2025.
24. H. Tella A. Hussei, S. Rc, B. Liu, A. Balghonaim , M. Mohandes, "Solar photovoltaic panel cells defects classification using deep learning ensemble methods," Case Stud. in Ther.Eng., vol. 66, p. 105749, 2025.
25. R. Priyadarshini, P. Manoharan, and S. Roomi, "EfficientNet-based deep learning for visual fault detection in solar photovoltaic modules," Tehnički vjesnik, vol. 32, no. 1, pp. 233–241, 2025.
26. S. Ahmed, H. Rashid, Z. Qadir, and Q. Tayyab, "Deep learning-based recognition and classification of soiled photovoltaic modules using HALCON software for solar cleaning robots," Sensors, vol. 25, no. 5, p. 1295, 2025.
27. P. K. Balachandran, M. A. A. M. Zainuri, and F. Alsaif, "Solar FaultNet: Advanced fault detection and classification in solar PV systems using SwinProba‐GeNet and BaBa optimizer models," Ener. Science & Eng., 2025.
28. B. Wang, Q. Chen, M. Wang, Yuntian Chen, Z. Zhang, X. Liu, W. Gao, Y. Zhang, and H. Zhang, "PVF-10: A high-resolution unmanned aerial vehicle thermal infrared image dataset for fine-grained photovoltaic fault classification," Applied Ener., vol. 376, p. 124187, 2024
29. T. Tuncer, S. Dogan, M. Baygin, I. Tasci, B. Mungen, B. Tasci, P.D. Barua, and U.R. Acharya, "Directed Lobish-based explainable feature engineering model with TTPat and CWINCA for EEG artifact classification," Know.-Based Syst., vol. 305, p. 112555, 2024.
30. J. Goldberger, G. E. Hinton, S. Roweis, and R. R. Salakhutdinov, "Neighbourhood components analysis," in Advances in Neural Information Processing Systems, vol. 17, 2004.
31. V. Vapnik, “The Nature of Statistical Learning Theory”. New York, NY, USA: Springer Science & Business Media, 2013.