Hourly Solar Irradiation Prediction by Artificial Neural Network Based on Similarity Analysis of Time Series

Abstract

Solar energy is a renewable energy source which has intermittent and variable characteristic. Solar irradiation must be predicted in advance in an electrical grid which has solar energy based electrical power generation systems in order to operate the electrical grid stable and efficient. In this study, a multi-layered, feed forward artificial neural network (ANN) has been designed to predict the hourly solar irradiation of next day. The designed ANN has been trained by data which has been obtained via similarity analysis. Total solar irradiation on horizontal plane, relative humidity and temperature data of Trabzon province for 2015-2017 have been used as the training data set. Hourly solar irradiation prediction has been performed by utilizing the designed ANN and test data set. The prediction results have been evaluated as to root mean square (RMS), mean absolute error (MAE) and mean absolute percentage error (MAPE) performance criteria. The obtained performance criteria results show that the proposed ANN could make prediction with acceptable error.