基于EWT-FIG和ORELM模型的风速多步区间预测

针对风速预测具有较强的不确定性,提出了一种经验小波变换一模糊信息粒化和变异鲁棒极限学习机组成的短期风速区间预测模型。该模型采用经验小波变换将原始风速分解为若千个模态分量和一个剩余量,并对所有分量进行重构,为了缩小预测区间范围,仅对重构后的剩余量进行模糊粒化,根据需求提取每个窗口的最大值、平均值和最小值,然后对极限学习机进行优化,最后对所有分量建立离群鲁棒极限学习机预测模型,叠加预测值实现风速多步区间预测。实际算例表明:所提多步区间预测方法能有效跟踪风速变化,具有较高的预测精度和可靠的区间预测效果。

Wind speed multi - step interval prediction based onEWT- FIG and OREI M model

For the strong uncertainty of the wind speed forecasting, a short-term wind speed intervalforecasting model composed of empirical wavelet transform-fuzzy information granulation( EWT -FIG ) and outlier-robust extreme learning machine ( ORELM ) is proposed. The model uses EWT todecompose the original wind speed into several model components and a residual, and reconstructs allcomponents. Fuzzy information granulation of the reconstructed residual is made in order to narrow theprediction range. The minimum, average and maximum value of each window are extracted accordingto the need. Then, the extremelearning machine ( ELM) is optimized. Finally, the ORELM predic-tion model is established for all components, and the predicted value is superimposed to realize windspeed multi-step interval prediction. Practical examples show that the proposed multi-step intervalprediction method can effectively track changes in wind speed, and has higher prediction accuracyand reliable interval prediction effect.