Search

Wind energy is an important green energy. The use of wind energy can alleviate the pressure caused by the shortage of traditional energy. The wind speed is affected by many factors, which makes it difficult to forecast accurately. Most wind speed forecasting methods only consider the wind speed data. The other influence factors are usually ignored. In this paper, we use traditional wind speed data and weather factors as the input of our proposed model. In the proposed model, wavelet threshold is employed to reduce the noise of raw wind speed data. We apply grey relational analysis to select the weather factors that have great influence on wind speed. The wind speed data are the input vectors of Elman neural network (Elman) and back propagation optimized by cuckoo search. Weather factors are the input vectors of WNN-GRNN which combines the wavelet neural network (WNN) and general regression neural network (GRNN). Finally, the forecasting results of the weather factors and wind speed data are combined by variance reciprocal method. The data sets at Kalaeloa Oahu, Hawaii are chosen to test the validity of the proposed model. The results show that the proposed model has obvious advantages compared to other benchmark forecasting models. [ABSTRACT FROM AUTHOR]