Power Forecasting of a Photovoltaic Plant Located in ENEA Casaccia Research Center

This work proposes an Artificial Neural Network (ANN) able to provide accurate forecasting of power produced by photovoltaic (PV) plants. The ANN is customized on the basis particular season year. An analysis input variables, i.e., solar irradiance, temperature and air humidity, carried out means Pearson Correlation, has allowed select, day day, most suitable set inputs architecture also reduce necessity large computational resource. Thus, features are added as needed, avoiding waste resources. method been validated through data collected from a PV plant installed in ENEA (National agency for new technologies, energy sustainable economic development) Research Center, located Casaccia, Rome (Italy). developed strategy furnish predictions even case strong irregularities providing results rapidly changing scenarios.