Generation of Photovoltaic Output Power Forecast Using Artificial Neural Networks

An artificial neural network (ANN) model is used for forecasting the power provided by photovoltaic solar panels using feed forward neural network (FFNN) of a photovoltaic installation located in the city of Mohammedia (Morocco). One year of hourly data on solar irradiance, ambient temperature and output PV power were available for this study. For this, different combinations of inputs with different numbers of hidden neurons were considered. To evaluate this model several statistic parameters were used such as the coefficient of correlation (R), the Root Mean Squared Error (RMSE) and the Mean Absolute Error (MAE). The results of this model tested on unknown data showed that the model works well, with regression coefficients lying between 99.6% and 99.8% for sunny days and between 93% and 96% for cloudy days.