Analysis and validation of 24 hours ahead neural network forecasting of photovoltaic output power

• We propose an artificial neural network for photovoltaic energy forecasting.
• We analyze its sensitivity with respect to the input data sets and error definitions.
• Data are taken from experimental activities carried out on a real photovoltaic plant.
• The hourly energy prediction covers all the daylight hours of the following day.

In this paper an artificial neural network for photovoltaic plant energy forecasting is proposed and analyzed in terms of its sensitivity with respect to the input data sets.Furthermore, the accuracy of the method has been studied as a function of the training data sets and error definitions. The analysis is based on experimental activities carried out on a real photovoltaic power plant accompanied by clear sky model.In particular, this paper deals with the hourly energy prediction for all the daylight hours of the following day, based on 48 hours ahead weather forecast. This is very important due to the predictive features requested by smart grid application: renewable energy sources planning, in particular storage system sizing, and market of energy.