A Hybrid Clustering and Neural Network-based Ensemble Method for Day-Ahead PV Output Forecast

With the increasing demand for green energy, solar photovoltaic (PV) has become as a major source of electricity production. However, the high penetration of solar PV in the power grid poses challenges due to variations in electric power generation, making accurate PV output forecasting crucial for effective system operation. This paper proposes a hybrid method for day-ahead PV output forecasting based on historical data from an actual PV power plant. The study employs different weather clustering techniques and deep learning models. The forecast results of these models are then combined using an ensemble method. Results show that the proposed model improves accuracy across various weather conditions, and the method is suitable for day-ahead PV output forecasting.