Predicting power production from a photovoltaic panel through artificial neural networks using atmospheric indicators

Abstract

In this study, an artificial neural network was modeled in order to predict the power generated by a monocrystalline silicon photovoltaic panel. This experimental study measured and recorded the voltage and current generated by the photovoltaic panel for a year, along with environmental variables such as solar irradiance, air temperature, wind speed, wind direction, relative humidity, and angle of the sun's elevation. In the results of the comparisons between measured and estimated power, a perfect estimation was found to have been conducted in which the root mean square error did not exceed 1.4% and the coefficient of correlation (R) ranged from 99.637 to 99.998%. These results were obtained from the testing dataset. In this study, achieved artificial neural network models are able to perform estimations for any location using the atmospheric indicators. These models are considered able to lead investors using extremely sensitive and robust estimations in order to learn solar energy's potential in a location.