Open AccessNeural modelling of solar radiation variability
Author(s) -
T. Ronkiewicz,
Joanna Aleksiejuk-Gawron,
Michał Awtoniuk,
Jarosław Kurek
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1736/1/012015
Subject(s) - artificial neural network , total harmonic distortion , matlab , mean squared error , meteorology , computer science , sine wave , radiation , algorithm , machine learning , engineering , statistics , mathematics , geography , physics , electrical engineering , optics , voltage , operating system
Artificial neural networks are increasingly used in engineering and technical sciences, especially to solve problems under process uncertainty. The mathematical model presented in this article describes cloud variability. The application of the model can increase the efficiency of solar systems because the response time of the solar panel to changing weather conditions is crucial. The model involves an artificial neural network that serves to determine the degree of daily cloud coverage based on three data – the month, daily solar radiation sum and total harmonic distortion factor (THD). The THD factor is determined for daily solar radiation courses using a Fast Fourier Transform. Approaching the daily variability of solar radiation as a sine wave allows employing the THD factor in an unconventional and innovative way. The modelling data have been derived from the measurements of the meteorological station of the Institute of Mechanical Engineering of the Warsaw University of Life Sciences. MATLAB Software (2019a) was used for data processing and network modelling. The model is verified using the mean square error. The performed analysis provides promising results and conclusions.