基于改进的加权一阶局域预测模型的风电功率预测

Prediction of wind power based on improved weighted local-region multi-step forecasting model

为提高风电功率预测精度,将混沌时间序列分析应用于风电序列,研究风电数据混沌特性以及混沌时间序列加权一阶局域多步预测法(AOLMM).针对高维相空间中相点间的相关性大小不能被欧式距离精确反映的问题,利用灰色关联度和相点间的距离确定邻近点的权重大小,同时将中心点与邻近点延迟矢量最后一个分量的强相关性考虑在内,改进了预测方法.对风电功率预测分析可见,改进的方法具有较好的适应能力和预测精度.

To improve the accuracy of wind power time series short-term forecasting,the chaotic time series analysis is applied to wind power sequence;the chaotic characteristic of wind power time series and weighted one rank local multi-steps forecasting model(AOLMM)is discussed.The correlation of phase point is not reflected by the Euclid in high embedding dimension phase space,an improved weighted one rank local forecasting model is presented.The weights of grey relational degree and Euclidean distance for phase points are weighted and applied to chaotic local forecasting model;the strongest correlation of center point and the last component of neighborhood are taken into consideration.Through the analysis of the wind power time series prediction,the results show the model has better adaptability and prediction accuracy.