New approach on renewable energy solar power prediction in Indonesia based on Artificial Neural Network technique: Southern region of Sulawesi island study case

Abstract

Indonesia located at 94 o -141 o E and 6 o N -11 o S is the largest archipelago in the equator on earth. As a tropical country, Indonesia is endowed with abundant solar energy potential. This study is focused on modeling the Global Solar Radiation using Artificial Neural Network to predict GSR in a location which is available with meteorological data but lack with radiation measurement data. A case study on 5 locations in South Western region of Sulawesi was used to develop the model. The ANN model used 4 location with 5 years monthly meteorological and radiation data for training, and one location for testing. The simulation shows that an ANN with 4 layers and 5 neurons is the most appropriate model with an MSE of 0.003 and r of 0.99937. The model provides an excellent performance of prediction of with an MPE of 0.1427% and r 2 of 0.999967.The predicted radiation data is in reasonable agreement with the actual data at the testing location; this shows the ability of ANN technique in generalization of data unavailability and produces an accurate prediction