Investigation of the feasibility of wind speed prediction using hybrid model of neural networks, neural -fuzzy networks and wavelet (Case Study: Station of Yazd)

Prediction of wind components (like wind speed) is considered one of the important factor، especially in relation to evaporation in a watershed. In this paper، in order to increase the efficiency of artificial intelligence models، for predicting wind speed، two neural network and neuro-fuzzy models combined with wavelet and two new hybrid models were presented. This research was carried out by some climatic parameters in Yazd synoptic station including wind speed، mean temperature، maximum temperature، relative humidity and evaporation. Then، efficiency of wavelet neural network and wavelet neural –fuzzy models were compared with the neural network and neuro-fuzzy to predict the wind speed for next 12 months. Finally، in order to confirm the efficiency of the best model، wind speed in 2005 was predicted using effective climatic parameters in 2004. The obtained results showed higher efficiency of the wavelet neural network and wavelet neural – fuzzy than the neural network and neuro-fuzzy. Verification of hybrid models confirmed the efficiency of wavelet neural – fuzzy in comparing to the other models.