 |
|
|
Nurul Huda Ishak Electrical Engineering Studies, Universiti Teknologi MARA, Cawangan Pulau Pinang, Permatang Pauh Campus, 13500 Pulau Pinang, Malaysia Iza Sazanita Isa Electrical Engineering Studies, Universiti Teknologi MARA, Cawangan Pulau Pinang, Permatang Pauh Campus, 13500 Pulau Pinang, Malaysia Muhammad Khusairi Osman Electrical Engineering Studies, Universiti Teknologi MARA, Cawangan Pulau Pinang, Permatang Pauh Campus, 13500 Pulau Pinang, Malaysia Kamarulazhar Daud Electrical Engineering Studies, Universiti Teknologi MARA, Cawangan Pulau Pinang, Permatang Pauh Campus, 13500 Pulau Pinang, Malaysia Mohd Shawal Jadin Faculty of Electrical and Electronics Engineering Technology, Universiti Malaysia Pahang Al-Sultan Abdullah, 26600 Pekan, Pahang, Malaysia Noor Fadzilah Razali Electrical Engineering Studies, Universiti Teknologi MARA, Cawangan Pulau Pinang, Permatang Pauh Campus, 13500 Pulau Pinang, Malaysia
|
|
| Abstract | |
|
Intelligent diagnostic models that can identify and categorise thermal anomalies affecting efficiency and safety are essential for the reliable operation of photovoltaic (PV) systems. One of the most serious defects is thermal hotspots, which frequently result in energy loss and potential fire hazards. Convolutional Neural Networks (CNNs) have demonstrated promising performance when applied to aerial thermal images for PV fault detection. However, most existing models are limited to binary fault identification, thereby constraining their applicability for risk-based maintenance. Conventional CNN architectures also lack an optimised flatten-layer representation and fail to capture the full variability in hotspot severity. This study presents an enhanced CNN-based framework for multiclass classification of hotspot severity in PV modules. To improve feature abstraction and class separability across severity levels, the proposed model employs a staged architectural refinement strategy by progressively adding convolutional layers to a baseline CNN. Hotspot regions extracted from unmanned aerial vehicles (UAV)-acquired thermal imagery were categorised into low, medium, and high severity levels for model training and evaluation. The improved model achieved a 9.1% increase in accuracy and a 7.6% increase in F1-score, outperforming the baseline CNN, thus confirming its superior discriminative learning capability and diagnostic robustness. These findings demonstrate that architectural deepening and flatten-layer optimisation can advance automated thermographic inspection toward severity-aware predictive maintenance for PV systems. |
|
| Keyword: Photovoltaic, Convolutional Neural Networks, Hotspot, Severity,Classification | |
| References: | |
|
[1] M. I. Kolawole and B. L. Ayodele, “Smart electronics in solar-powered grid systems for enhanced renewable energy efficiency and reliability,” International Journal of Science and Research Archive, vol. 13, no. 2, pp. 2910–2930, 2024. Available: https://doi.org/ 10.30574/ijsra.2024.13.2.2512 [2] E. H. Sepulveda-Oviedo, “Impact of environmental factors on photovoltaic system performance degradation,” Energy Strategy Reviews, vol. 59, no. October 2024, p. 101682, 2025. Available: https://doi.org/10.1016/j.esr.2025.101682 [3] N. Prajapati, R. Aiyar, A. Raj, and M. Paraye, “Detection and Identification of faults in a PV Module using CNN based Algorithm,” 2022 3rd International Conference for Emerging Technology, INCET 2022, pp. 1–5, 2022. Available: https://doi.org/ 10.1109/INCET54531.2022.9825452 [4] N. Drir, A. Mellit, and M. Bettayeb, “A Novel Ensemble CNN Framework With Weighted Feature Fusion for Fault Diagnosis of Photovoltaic Modules Using Thermography Images,” IEEE J. Photovolt., vol. 15, no. 1, pp. 146–154, 2024. Available: https://doi.org/ 10.1109/JPHOTOV.2024.3492283 [5] W. M. Elmessery, A. Habib, M. Y. Shams, T. Abd El-Hafeez, and T. M. El-Messery, “Deep regression analysis for enhanced thermal control in photovoltaic energy systems,” Sci. Rep., vol. 14, no. 1, pp. 1–22, 2024. Available: https://doi.org/ 10.1038/s41598-024-81101-x [6] Q. Guo, B. Han, P. Chu, Y. Wan, and J. Zhang, “MF-FusionNet: A Lightweight Multimodal Network for Monitoring Drought Stress in Winter Wheat Based on Remote Sensing Imagery,” Agriculture (Switzerland), vol. 15, no. 15, pp. 1–28, 2025. Available: https://doi.org/ 10.3390/agriculture15151639 [7] M. Dhimish, M. Theristis, and V. d’Alessandro, “Photovoltaic hotspots: A mitigation technique and its thermal cycle,” Optik (Stuttg)., vol. 300, no. November 2023, p. 171627, 2024. Available: https://doi.org/ 10.1016/j.ijleo.2024.171627 [8] G. E. Mustafa Abro, A. Ali, S. Ali Memon, T. Din Memon, and F. Khan, “Strategies and Challenges for Unmanned Aerial Vehicle-Based Continuous Inspection and Predictive Maintenance of Solar Modules,” IEEE Access, vol. 12, no. December, pp. 176615–176629, 2024. Available: https://doi.org/10.1109/ACCESS.2024.3505754 [9] M. Aghaei, M. Kolahi, A. Nedaei, N. S. Venkatesh, and S. M. Esmailifar, “Autonomous Intelligent Monitoring of Photovoltaic Systems: An In-Depth Multidisciplinary Review,” Progress in Photovoltaics: Research and Applications, vol. 33, no. 3, pp. 381–409, 2025. Available: https://doi.org/10.1002/pip.3859 [10] V. K. Y. Swatilekha Ghosh, Santosh Kumar Singh, “Experimental investigation of hotspot phenomenon in PV arrays under mismatch conditions,” Solar Energy, vol. 253, 2023. Available: https://doi.org/10.1016/j.solener.2023.02.033 [11] O. E. Olabode, I. K. Okakwu, D. O. Akinyele, T. O. Ajewole, S. Oyelami, and O. V. Olisa, “Effect of Ambient Temperature and Solar Irradiance on Photovoltaic Modules’ Performance,” Iranica Journal of Energy and Environment, vol. 15, no. 4, pp. 402–420, 2024. Available: https://doi.org/ 10.5829/ijee.2024.15.04.08 [12] A. Raorane, D. Magare, and Y. Mistry, “Thermal image-driven CNN for predicting solar photovoltaic module lifespan from hotspots,” Solar Energy, vol. 302, no. July, p. 113965, 2025. Available: https://doi.org/ 10.1016/j.solener.2025.113965 [13] A. K. Vahid Sinap, “CNN-based automatic detection of photovoltaic solar module anomalies in infrared images: a comparative study,” Neural Comput. Appl., 2024. Available: https://doi.org/10.1007/s00521-024-10322-y [14] L. Zhang, J. Kuang, Y. Teng, S. Xiang, L. Li, and Y. Zhou, “A Lightweight Infrared and Visible Light Multimodal Fusion Method for Object Detection in Power Inspection,” Processes, vol. 13, no. 9, pp. 1–18, 2025. Available: https://doi.org/10.3390/pr13092720 [15] Z. Barraz, I. Sebari, H. Oufettoul, K. Ait el kadi, N. Lamrini, and I. Ait Abdelmoula, “A holistic multimodal approach for real-time anomaly detection and classification in large-scale photovoltaic plants,” Energy and AI, vol. 21, no. May, p. 100525, 2025. Available: https://doi.org/ 10.1016/j.egyai.2025.100525 [16] M. Z. A. Hamid, K. Daud, Z. H. C. Soh, M. K. Osman, I. S. Isa, and N. H. Ishak, “Review of Deep Learning-Based Hotspot Detection in Solar Photovoltaic Arrays,” 2024 IEEE 4th International Conference in Power Engineering Applications: Powering the Future: Innovations for Sustainable Development, ICPEA 2024, no. March, pp. 332–337, 2024. Available: https://doi.org/ 10.1109/ICPEA60617.2024.10498989 [17] M. U. Ali, A. Zafar, W. Ahmed, M. Aslam, and S. H. Kim, “Enhancing Photovoltaic Reliability: A Global and Local Feature Selection Approach with Improved Harris Hawks Optimization for Efficient Hotspot Detection Using Infrared Imaging,” Int. J. Energy Res., vol. 2024, no. 1, 2024. Available: https://doi.org/10.1155/2024/5586605 [18] E. Tarigan, “Identification of early operational defects in photovoltaic modules: A case study of a 24.9 MWp solar PV system in Sumatra, Indonesia,” Unconventional Resources, vol. 6, no. February, p. 100156, 2025. Available: https://doi.org/10.1016/j.uncres.2025.100156 [19] K. Masita, A. Hasan, T. Shongwe, and H. A. Hilal, “Deep learning in defects detection of PV modules: A review,” Solar Energy Advances, vol. 5, no. September 2024, p. 100090, 2025. Available: https://doi.org/10.1016/j.seja.2025.100090 [20] D. Gertsvolf, M. Horvat, D. Aslam, A. Khademi, and U. Berardi, “A U-net convolutional neural network deep learning model application for identification of energy loss in infrared thermographic images,” Appl. Energy, vol. 360, no. January, p. 122696, 2024. Available: https://doi.org/ 10.1016/j.apenergy.2024.122696 [21] Y. Zefri, I. Sebari, H. Hajji, and G. Aniba, “An Encoder-Decoder U-Net Based Model for Overheated Photovoltaic Modules Extraction from Orthorectified Remotely Sensed Thermal Infrared UAV Imagery,” 2022 IEEE Mediterranean and Middle-East Geoscience and Remote Sensing Symposium, M2GARSS 2022 - Proceedings, pp. 25–28, 2022. Available: https://doi.org/ 10.1109/M2GARSS52314.2022.9840254 [22] M. Abdelsattar, A. AbdelMoety, and A. Emad-Eldeen, “ResNet-based image processing approach for precise detection of cracks in photovoltaic panels,” Sci. Rep., vol. 15, no. 1, pp. 1–22, 2025. Available: https://doi.org/10.1038/s41598-025-09101-z [23] Z. Li, W. Liu, G. Wu, and S. Yang, “Semantic segmentation of unmanned aerial vehicle image based on Resnet-Unet,” 2023 8th International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS), vol. 8, pp. 295–299, 2023. Available: https://doi.org/10.1109/iciibms60103.2023.10347663 [24] G. Mannino et al., “Photovoltaic Module Degradation Forecast Models for Onshore and Offshore Floating Systems,” Energies (Basel)., vol. 16, no. 5, 2023. Available: https://doi.org/ 10.3390/en16052117 [25] M. Shafiq and Z. Gu, “Deep Residual Learning for Image Recognition: A Survey,” Applied Sciences (Switzerland), vol. 12, no. 18, pp. 1–43, 2022. Available: https://doi.org/ 10.3390/app12188972 [26] M. Z. A. Hamid, K. Daud, Z. H. C. Soh, M. K. Osman, and I. S. Isa, “Deep learning-driven thermal imaging hotspot detection in solar photovoltaic arrays using YOLOv10,” 2024. Available: https://doi.org/10.24191/esteem.v20iSeptember.18 [27] B. Eri, “Early Hotspot Detection in Photovoltaic Modules using Deep Learning Methods,” vol. 09, no. 09, pp. 5065–5072, 2024. Available: https://doi.org/10.47191/etj/v9i09.10 [28] S. Goyal and J. C. Rajapakse, “Self-supervised learning for hotspot detection and isolation from thermal images,” Expert Syst. Appl., vol. 237, no. PB, p. 121566, 2024. Available: https://doi.org/ 10.1016/j.eswa.2023.121566 [29] U. Pruthviraj, Y. Kashyap, E. Baxevanaki, and P. Kosmopoulos, “Solar Photovoltaic Hotspot Inspection Using Unmanned Aerial Vehicle Thermal Images at a Solar Field in South India,” Remote Sens. (Basel)., vol. 15, no. 7, 2023. Available: https://doi.org/10.3390/rs15071914 [30] L. Alzubaidi, J. Zhang, A. J. Humaidi, A. Al-Dujaili, and Y. Duan, “Review of deep learning: concepts, CNN architectures, challenges, applications, future directions,” J. Big Data, vol. 8, no. 1, 2021. Available: https://doi.org/10.1186/s40537-021-00444-8
|
|
Performance improvement of CNN-based model for multiclass hotspot severity classification in photovoltaic thermal imagery
