Modeling and Prediction of Photovoltaic Power Output Using Artificial Neural Networks Considering Ambient Conditions

Abstract

The forecasting of photovoltaic power output with a reliable predictive tools considering ambient conditions is very important in order to dissemination the technologies of the PV system, as well , to improve the performance of the photovoltaic systems in the planning. The aim of this work is present a solar power modeling method using artificial neural networks (ANNs). Two neural network structures, namely, general regression neural network (GRNN) feed-forward back propagation (FFBP), have been used to model a photovoltaic output power and approximate the generated power. Both neural networks have five inputs and one output. The inputs are ambient temperature, cell temperature, wind speed, humidity, and irradiance; the output is the power. The data used in this paper taken from the experimental work which has been conducted in the energy laboratory of energy engineering department, Baghdad city, started from January, 2017 to May, 2017. The five months of data was used for training and testing the neural networks. The results show that, the solar irradiance has the greatest effect on the estimation of the photovoltaic power output with ratio (33.7%) then the cell temperature, and ambient temperature with ratios (28.5%, 25.3%), respectively, while humidity has medium effects with ratio (12.4%). Wind speed has the least effect with ratio (0.1%). The results of simulation indicate that the two models were accurate and can be effectively used for prediction of photovoltaic power output. However, it was demonstrated that the GRNN network model gave more accurate results when compared with those obtained using FFBP network model. As well, GRNN network proved its ability to predict in less time than FFBP network.