The Use of Meteorology Data in Short-Term Prediction of Wind Speed for Wind Turbine Using Elman Recurrent Neural Network

Abstract

Wind energy is one of the promising renewable energy sources that are ideal for daily use, especially in the area with sufficient wind blows like Indonesia. Wind speed caused by wind energy is a driving force for wind turbines to produce electrical power. One problem in wind turbine management is to predict the speed of the wind in the short term for efficiency. In this research, forecasting of short-term wind speed was done in the city of Sibolga by uses an Elman recurrent neural network based on meteorological data: temperature, humidity, and air pressure to predict over the next ten days. Four prediction models were developed for this purpose based on training parameters and dataset used. The wind speed forecasting produces MAPE error values of 20.02% in the first model, 23.31% in the second model, 18.15% in the third model, and 12.51% in the fourth model. The fourth model was capable of predicting with the lowest error and, therefore, considered to be useful for wind turbine management.