Global Solar Radiation Forecasting with Artificial Neural Networks

Abstract

This study presents a detailed examination of using artificial neural networks for predicting global solar radiation. The research aims to develop an artificial neural network model using five years of solar radiation and meteorological variables (precipitation, wind speed, relative humidity, vapor pressure, cloudiness, current pressure, average temperature, number of sunny days, solar radiation, and daily average solar intensity) obtained from the central meteorological observation station of Kocaeli province between 2017 and 2021. The model aims to address the complexity of solar radiation as a phenomenon and the challenges associated with direct measurement. Artificial neural networks are considered an ideal tool for this purpose due to their ability to analyze complex data structures and identify relationships. The dataset used in this study includes detailed measurements of five years of solar radiation and meteorological variables collected from the meteorological observation station. These data encompass factors crucial for solar radiation prediction and provide information to enhance the accuracy of the model. The dataset is divided into training, validation, and testing phases, and relevant metrics are used to evaluate the performance of the artificial neural network model. The results demonstrate the successful prediction of global solar radiation by the developed artificial neural network model. The model undergoes a learning process to comprehend the complexity of solar radiation and make predictions by utilizing the relationships between various meteorological variables. This study emphasizes the importance of solar radiation prediction in areas such as solar energy projects, energy planning, and climate change research.