Wind speed forecasting using FEEMD echo state networks with RELM in Hebei, China

Reducing the dependence on fossil-fuel-based resources is becoming significant due to the detrimental effects on environment and global energy-dependent. Thus, increased attention has been paid to wind power, a type of clean and renewable energy. However, owing to the stochastic nature of wind speed, it is essential to build a wind speed forecasting model with high-precision for wind power utilization. Therefore, this paper proposes a hybrid model which combines fast ensemble empirical model decomposition (FEEMD) with regularized extreme learning machine (RELM). The original wind speed series are first decomposed into a limited number of intrinsic mode functions (IMFs) and one residual series. Then RELM is built to forecast the sub-series. Partial auto correlation function (PACF) is applied to analyze the intrinsic relationships between the historical speeds so as to select the inputs of RELM. To verify the developed models, short-term wind speed data in July 2010 and monthly data from January 2000 to May 2010 in Hong songwa wind farm, Chengde city are used for model construction and testing. Two additional forecasting cases in Hebei province are also applied to prove the model's validity. The simulation test results show that the built model is effective, efficient and practicable.