Application of a Data-Driven Anomaly Detection Approach for a Solar Photovoltaic Plant Using Real-Time SCADA Data

This work presents a real-time approach to detect underperformances within a solar PV plant using real-time data collected from a Supervisory Control And Data Acquisition (SCADA) system. The approach is based on the combination of a physical and a statistical additive model to simulate how much energy will be generated from the PV systems. The comparison of the models results to the measured data is used to spot power curtailment and underperformance issues in real-time. Historical measurements are used to develop and tune the models. The proposed approach is evaluated on real PV power generation data, from several months of data measurements of multiple PV systems ranging from 9 to 16 kWp of different PV technologies (m-Si, P-Si, CIS) installed in the solar test facility of Green Energy Park in Morocco. In order to facilitate the implementation of the model in an embedded platform, the simplicity of the model and the lower computational time and cost were the main drivers for this research.