Feasibility of soft computing techniques for estimating the long-term mean monthly wind speed

Abstract

Estimating wind energy plays an important role in energy science as it can be considered a crucial source of renewable and sustainable energy. In this study, five types of soft computing approaches were implemented to estimate the long-term mean monthly wind speed (W) at 50 weather stations in Iran. The applied models were artificial neural networks (ANN), gene expression programming (GEP), multivariate adaptive regression spline (MARS), adaptive neuro-fuzzy inference system (ANFIS), and random forest (R.F.). In addition, the geographical information (i.e., latitude, longitude, and altitude) and periodicity term (i.e., the number of months in a year) were used to input the models. Results demonstrated that the R.F. technique was the best model for estimating W, utilizing the geographical information and number of the month. Hence, it can be concluded that the applied soft computing techniques can employ the aforementioned inputs for estimating W.