“…The raised crucial problem is the strong dependence of the system response from many extrinsic factors, such as insolation intensity, ambient temperature, cell temperature, air velocity, humidity, cloudiness and pollution. All these factors have to be taken into account for the modelling of a PV plant [1]- [3], for the tracking of the Maximum Power Point (MPP) [4], [5], for the monitoring of the energy performance [6], [7], for the analysis and the modelling of the defects [8], [9], for the planning of the day after, for the forecasting purposes [10]- [13]. As reported in [13], PV generation forecasting methods can be broadly classified into three approaches: a) the Numerical Weather Prediction (NWP)-based forecast [14], which uses the first principles for predicting solar irradiance and PV generation; b) the data-driven statistical approach [15], [16], which includes Auto-Regression (AR)-based models and computational intelligence tools such as Artificial Neural Networks (ANNs) [9]; c) the hybrid one, which combines the NWP-based and data-driven models.…”