Day Ahead Solar Irradiation Forecasting Based on Extreme Learning Machine

Abstract

Solar radiation data is very important for humans in meteorology, agriculture and energy. An Extreme Learning Machine (ELM) model is a data-based model developed from a single hidden layer feed-forward neural network (SLFN) which has the superiority in terms of training speed that is better than its predecessor generation. A model for predicting solar radiation in the Manokwari area and its surroundings was built with the ELM algorithm. The model has been used to predict daily solar radiation in the area. The ELM model has been trained using 8016 data solar irradiation and temperature from NASA. The test results show that the built has fairly high accuracy with MAE values of about 0.6392 in a training time of 4.4375 seconds. The ELM model has superiority in time consuming compared to a simple feedforward neural network.