부산대학교 도서관

Non-conventional energy resources in which wind farms produced power treated as important substitutes in power system networks including their suitable atmospheric effects. Forecasting (short-term) of wind speed has large impact for taking decisions in load variations as well as economic load dispatch within wind integration based power systems. The nature of wind power is stochastic and intermittent. Wind power is not transferable every times because it depends on various atmospheric conditions, so accurate prediction is needed. ANN algorithms in which Levenberg-Marquardt back propagation, Scaled Conjugate Gradient algorithm along with Bayesian Regularization are applied for the forecasting of wind speed on short-term basis which indicates one hour ahead forecasting of wind speed for Knock Airport, Ireland on hourly pattern with help of MATLAB R2014a. Hourly pattern historical data of temperature, wind speed and its direction are adapted for the performing of forecasting. Results of simulation represent very precise one hour ahead forecasting of wind speed with less error.