Solar production forecasting based on irradiance forecasting using artificial neural networks

There is a growing awareness that forecasting of solar irradiance is of special importance for forecasting the power output of photovoltaic (PV) systems and thus for optimizing their operation. This work presents the development of solar irradiance and PV power output forecasting models, based on artificial neural networks (ANNs), operating with a time horizon of 24 h in order to be integrated as part of home energy management systems (HEMS). The key characteristic of the proposed approach consists of employing statistical feature parameters to reduce the size of input data, while the results obtained indicate that it provides a reasonable balance between computational requirements and forecasting accuracy of the PV power output within the considered time frame.