Open AccessPrediction of Day Ahead Hourly Solar Radiation by Meteorological Forecasting Supported Artificial Neural Network: A Case Study for Trabzon Province
Author(s) -
Sibel Cevik,
Recep Çakmak,
İsmail H. Altaş
Publication year - 2018
Publication title -
akıllı sistemler ve uygulamaları dergisi
Language(s) - English
Resource type - Journals
ISSN - 2667-6893
DOI - 10.54856/jiswa.201812029
Subject(s) - mean absolute percentage error , mean squared error , artificial neural network , renewable energy , context (archaeology) , meteorology , correlation coefficient , computer science , environmental science , simulation , statistics , machine learning , engineering , mathematics , geography , archaeology , electrical engineering
Electricity generation from renewable energy sources is increased day by day. Accurate estimation of electricity generation from the renewable energy sources which have intermittent and variable characteristics is a requirement to ensure stable operation of the electrical grid. In this study, a multi-layer artificial neural network (ANN) system, which is supported by meteorological forecasting data, has been proposed to predict day ahead hourly solar radiation. In this context, the ANN system which operates by based on cause-effect relationship has been designed. In order to increase accuracy of the solar radiation prediction of the designed ANN, a similar day selection algorithm has been developed. A unique ANN has been constituted for each season by evaluating the seasons within itself. The designed ANN model has been designed, trained and tested in MATLAB simulation environment without using codes of the MATLAB ANN toolbox. Day ahead hourly solar radiation of Trabzon province has been predicted by the proposed ANN. The accuracy of the predictions has been evaluated by the mean absolute percentage error (MAPE), the root means squared error (RMSE), the mean absolute error (MAE) and the correlation coefficient (r) performance measures.