A Novel Online Method for Wind Speed Forecasting Based on Hybrid Decomposition and OSORELM

Accurate forecasting of wind speed is crucial for wind power management and power system operation. In this paper, a novel online method based on hybrid decomposition and Online Sequential Outlier Robust Extreme Learning Machine (OSORELM) is proposed for wind speed prediction in a multi-step fashion. The hybrid decomposition method is composed of variational mode decomposition algorithm, sample entropy approach and wavelet packet decomposition algorithm (VMD-SE-WPD), and the OSORELM is optimized by Crisscross optimization (CSO). In data pre-processing period, the wind speed is decomposed by the proposed hybrid VMD-SE-WPD, where the VMD is first applied to decompose the original wind speed into a set of sub-series. Then, all sub-series are analyzed by SE method, and the part of sub-series with high SE value is further decomposed by WPD. Lastly, all sub-series obtained from above are forecasted by OSORELM, and the results of all sub-series are generated to obtain the final prediction results of the wind speed. The experiment results show that the proposed forecasting model can greatly improve the performance of wind speed prediction.