Predictive Analytics for Maximizing the Photovoltaic System Performance

• Published in: Conference record of the IEEE Photovoltaic Specialists Conference pp. 0210 - 0213
• Main Authors: Livera, Andreas, Marangis, Demetris, Tziolis, Georgios, Paraskeva, Vasiliki, Makrides, George, Georghiou, George E.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 09.06.2024
• Subjects: Machine learning algorithms; Maintenance; Photovoltaic systems; Prediction algorithms; Predictive analytics; Predictive maintenance; Predictive models; Schedules; Software; Software algorithms
• ISSN: 2995-1755
• DOI: 10.1109/PVSC57443.2024.10749069

To ensure the smooth Terawatt (TW) photovoltaic (PV) transition and secure the TWh power generation over the service lifetime, enhanced digitalization, and automation approaches for effective operation and maintenance (O&M) of PV systems are required. This highlights the urgent need for corrective and predictive analytics to detect and predict possible anomalies in PV systems and to effectively manage and optimally schedule the field O&M activities. To fulfill this need, a failure detection algorithm along with a predictive maintenance model, that leverages statistical and machine learning (ML) principles, are proposed to accurately detect failures and to predict potential fault issues before they occur. The developed algorithm and the model were benchmarked using historical field data from a PV power plant in Greece. The obtained results showed the effectiveness of the algorithm for accurate failure detection and the capability of the predictive model for generating predictive maintenance alarms. The integration of corrective and predictive analytics into a monitoring software solution can help PV plant operators to schedule the field O&M activities, by providing valuable fault insights and actionable maintenance recommendations.