Hybrid Prediction Model for Ultra-short-term Wind Speed Based on Empirical Mode Decomposition
As the issue of carbon emissions becomes increasingly urgent,wind energy,as a clean and renewable energy source,has attracted widespread attention and research.However,due to the uncertainty and intermittence of wind,the utilization rate of wind energy is low and its energy efficiency cannot be fully exerted.Therefore,we propose a line-ar and nonlinear hybrid model based on empirical mode decomposition(LNH-EMD)prediction approach to forecast ul-tra-short-term wind speed according to the chaotic characteristics of wind.Firstly,we decompose the historical wind speed series into several groups of relatively simple modal components by EMD.Secondly,we performe a linear judg-ment of each component separately.We complete the prediction proceeding of nonlinear modal components based on the phase-space reconstruction,and use OLS fitting to forecast other modal components.Finally,we obtain the wind speed prediction results through empirical mode reconstruction of each component.The prediction results of typical nonlinear signals and real wind speed data from NERL illustrate that:the proposed LNH-EMD wind speed prediction model ex-hibits higher prediction accuracy and operation efficiency when compared to Gray,WA and LSTM prediction models.
ultra-short-term predictionwind speed predictionEMDhybrid modelphase space reconstruction
万方数据
维普
