A novel hybrid approach for wind speed prediction

Abstract

It is important to improve the accuracy of wind speed predictions for wind park management and for conversion of wind power to electricity. However, due to the chaotic and intrinsic complexity of weather parameters, the prediction of wind speed data using different patterns is difficult. A hybrid model known as SAM–ESM–RBFN is proposed for capturing these different patterns and obtaining better prediction performance. This model is based on the seasonal adjustment method (SAM), exponential smoothing method (ESM), and radial basis function neural network (RBFN). The mean hourly wind speed data from two meteorological stations in the Hexi Corridor of China were used as examples to evaluate the performance of the proposed approach. To avoid randomness due to the RBFN model or the RBFN component of the hybrid model, all of the simulations were repeated 30 times prior to averaging. The SAM–ESM–RBFN model numerically outperformed the following models: the Holt–Winters model (HWM), the multilayer perceptron neural network (MLP), the ESM, the RBFN, the hybrid SAM and ESM (SAM–ESM), the hybrid SAM and RBFN (SAM–RBFN), and the hybrid ESM and RBFN (ESM–RBFN). Overall, the proposed approach was effective in improving the prediction accuracy.