DWT-based Wind Speed Forecasting Using Artificial Neural Networks in the region of Annaba

Abstract

Global demand for electrical energy is in constant increase all over the world, leading to new kinds of energy from renewable resources, namely solar and wind power. Consequently, precise wind prediction is very important for efficient management of grid-connected wind farms. This article examines the use of wavelet analysis combined with neural networks to predict wind speed. The wavelet transform is employed to smooth the wind speed time series for better prediction using neural networks. Using the wind speed data of the region of Annaba situated in the east of Algeria, the obtained results show the db4 wavelet with 5-level decomposition outperforms all other wavelet families in terms of forecasting accuracy.