SCADA-driven variable similarity-based model for fault detection and predictive maintenance in photovoltaic systems

Abstrak

Global efforts to mitigate the rise in average global temperatures have increasingly emphasized the adoption of renewable energy sources. Among these initiatives, the deployment of solar power plants has emerged as a promising solution, utilizing photovoltaic (PV) systems to convert solar energy—an abundant, cost-free, and year-round resource—into electricity. Solar power plants present a viable alternative to fossil fuels traditionally used in thermal power generation. The reliability of PV systems, which directly influences their performance, functionality, safety, and economic viability, is a critical factor in realizing this potential. This article presents the development of a predictive maintenance framework for PV systems, incorporating anomaly detection and fault diagnosis based on supervisory control and data acquisition (SCADA) data. The methodology employs a variable similarity-based model (VBM) to identify anomalies and diagnose faults, while generating predictive alerts to inform operators of potential issues, thereby enabling proactive maintenance scheduling. The proposed framework is validated using real SCADA data collected from an operational solar power plant. The results demonstrate that the method effectively detects anomalies with reasonable accuracy, underscoring its practicality for application in solar power plant operations.