AI-DRIVEN PREDICTIVE MAINTENANCE FOR SOLAR PV PLANTS

Abstract

Utility-scale PV performance is increasingly determined not by nominal equipment ratings but by how quickly latent defects are detected and mitigated. In real operation, losses often emerge from combinations of minor issues: local hotspot development, connector degradation, string mismatch, elevated contact resistance, partial shading, and non-uniform soiling. When these conditions are not identified early, they accumulate and evolve into costly failure scenarios.

References

1. Mellit, A., & Kalogirou, S. (2022). Machine learning in photovoltaic systems: A review. Renewable Energy, 191, 447-472. https://doi.org/10.1016/j.renene.2022.06.105
2. PV Performance Modeling Collaborative. (2026). PVPMC resources and best practices. Sandia National Laboratories. https://pvpmc.sandia.gov/
3. Vasi, J., et al. (2020). The 2020 photovoltaic technologies roadmap. Journal of Physics D: Applied Physics, 53(19), 193001. https://doi.org/10.1088/1361-6463/ab9c6a
4. Park, N.-G., Gratzel, M., Miyasaka, T., Zhu, K., & Emery, K. (2019). Towards stable and commercially available perovskite solar cells. Nature Energy, 4, 669-679. https://doi.org/10.1038/s41560-019-0456-x
5. Fraunhofer Institute for Solar Energy Systems ISE. (2026). Photovoltaics Report. https://www.ise.fraunhofer.de/en/publications/studies/photovoltaics-report.html
6. International Energy Agency. (2024). Renewables 2024. https://www.iea.org/reports/renewables-2024