This study presents an innovative prediction method that integrates the single-diode model (SDM) and artificial neural network (ANN) to forecast the electrical characteristics and maximum power point of photovoltaic (PV) modules operating outdoors under varying PV module temperatures and solar radiation levels. A new approach based on the reduced form (RF) technique and particle swarm optimization (PSO) has been used to extract the estimated parameter variations of the series resistance and ideality factor during a reference day. These variations, along with photovoltaic module temperature and solar radiation, train an ANN model to predict these parameters for arbitrary days. The saturation current, shunt resistance, and photo-current are predicted using equations derived from the two previously predicted parameters and three key points of the electrical characteristics. Five predicted parameters of the SDM model are used to predict the current-voltage characteristics and maximum power. Validation using data from three PV technologies under different operating conditions and at geographically distinct sites from the National Renewable Energy Laboratory (NREL) demonstrates the accuracy of the proposed method, with mean normalized errors not exceeding 1.4338 %. • Improved forecasting of PV module electrical output under real-world conditions • Integration of the single-diode model with artificial neural networks • A new method for extracting the temporal variation of series resistance and ideality factor • Validation of maximum power prediction accuracy across various PV technologies • The mean normalized error doesn’t exceed 1.4338 %
Salah et al. (Tue,) studied this question.