A data approximation based approach to photovoltaic system maintenance

The solar panel, which transforms the energy car- ried by the light in electricity, is a reliable component of a photovoltaic (PV) system, but its efficiency depends on several factors, such as its orientation, its working temperature, and its tidiness. Since maintenance is an expensive activity, a careful evaluation of the degradation of the panel and the resulting production loss has to be carried out. Besides, an accurate estimation of the potential production with respect to the weather condition requires expensive instruments and skilled operators. In this paper, we propose an alternative approach based on the prediction of the potential production based on a public weather station in the nearby of the considered plant. Several computational intelligence paradigms as well as several prediction setups are here challenged and compared.