Performance Prediction of a Grid-Connected Photovoltaic Plant

Performance prediction issue is becoming more and more important for grid-connected Photovoltaic plants. This paper presents the performance prediction of a grid-connected photovoltaic plant using a hybrid technique; the used technique combines Seasonal Autoregressive Integrated Moving average (SARIMA) with Support Vector Machines (SVM). The employed data in this study are derived from a 20kWp grid-connected PV system installed on the rooftop of the municipality of Trieste (Italy). The results confirm the ability of the used SARIMA-SVM model to estimate the performance parameters with good accuracy.