Abstract：Aiming at the nonlinear and nonstationary of wind speed sequences, a novel method based on empirical mode decomposition (EMD) and radial basis function neural network (RBFNN) is proposed to improve the precision of short-term wind speed forecasting. The wind speed data is decomposed into a series of intrinsic mode function (IMF) components with similar time-frequency characteristics and stationary by using EMD to achieve the stationary of the wind speed data. The IMF components are predicted by RBFNN based on the time-frequency characteristics of different IMF components. The orthogonal least squares (OLS) is adopted to minimize the error rate. Finally, the each prediction results of IMF-RBFNN are restructured to obtain the last prediction result. The short-term wind speed forecasting system based on interactive GUI is designed and implemented. Forecasting results show that the EMD-RBFNN combined model can improve the forecasting accuracy of short term wind speed and is of a certain practical value.
2014, Vol. 15 
