The electricity grid is evolving to a distributed infrastructure in which smart grids integrating renewable energies will become dominant. Because of the limited capacity of the battery to store the energy produced at certain time of the day, it is necessary to shift the consumption to when the electricity is actually produced. This paper deals with the estimation of solar panel production in order to forecast when and how much electricity will be available. We propose an Artificial Neural Network model to predict the hourly production of photovoltaic (PV) plants. We evaluate our approach over a large dataset of solar panel electricity production over a period of seven years.