Prediction of Wind Speed with optimized EMD based RVFLN

Abstract

In this work, a new hybrid short-term wind speed forecasting approach which is the combination of Empirical mode decomposition (EMD) and Random vector functional link network (RVFLN) is proposed. In the beginning EMD is applied to decompose the chaotic historical wind speed data into number of intrinsic mode function (IMFs). These IMFs are passes through the proposed prediction model called as RVFLN (without weight optimization and with weight optimization (Levenberg-Marquardt algorithm (LMA)). The prediction model performance is tested with thirty minutes and one hour ahead wind speed data. The prediction performance of various prediction models are evaluated using three measure indices (Mean absolute percentage error (MAPE), Root mean square error (RMSE) and mean absolute error (MAE)).It is observed from the result that optimize EMD based RVFL giving better result against the EMD based RVFL without optimization, which proves the efficacy of the presented method for short-term wind speed forecasting. The overall results provided in the result section are satisfactory. Therefore the proposed LMA-EMD-RVFL is a very good forecasting algorithm for real time application in power system.