Wind Speed forecasting using empirical mode decomposition with ANN and ARIMA models

Abstract

Wind power output mainly depends on wind speed. Forecasting of wind speed is important for unit commitment, economic load dispatch planning, turbine active control and optimal planning for wind farms maintenance. In this paper wind speed has been forecasted for 30 hour ahead by using Artificial Neural Network (ANN) and Auto Regressive Integrated Moving Average (ARIMA) models based on Empirical Mode Decomposition (EMD) method. Wind speed data is decomposed into Intrinsic Mode Functions (IMF) and Residue by EMD method. High frequency IMFs are forecasted using ANN model and low frequency IMFs and a residue are forecasted using ARIMA model. The result obtained by proposed method has given less mean absolute percentage error (MAPE) and improved statistical parameters. Wind speed data of the site 7263 in the Midwest ISO region is used for this study and it has been taken from National Renewable Energy Laboratory (NREL) website for the year 2014.