Adaptive neuro-fuzzy short-term wind-speed forecasting for Egypt's East-Coast

Abstract

AbstractUse of wind energy as a renewable source of energy for electric utility systems is increasing around the world. The major challenges of wind energy generation are natural intermittency, unpredictability, and uncertainty due to wind variations. In this paper, five different adaptive neuro-fuzzy wind predictors are proposed and compared to forecast the speed of wind blowing in the East Coast of Egypt, a very promising location to generate more than 20 GW of wind power. The first and second proposed models are based on real wind-speed data of the selected site for the same month aided by the corresponding real wind-speed data for the same site for the past 4 and 6 years, respectively. The results that have been obtained from these models show more accuracy with respect to the previous work in the literature. Furthermore, three new models are proposed based on only 20% of the real data used for the first model obtaining similar accuracy to predict the average wind speed one day, half a day, and quarter day ahead.Keywords: adaptive neuro-fuzzy inference system (ANFIS)membership functionpower coefficienttip speed ratio