基于符号时间序列法的风电功率波动分析与预测

Analysis and Forecast of Wind Power Fluctuation Based on Symbolized Time Series Theory

风电功率波动与预测是风电并网研究的主要内容。针对风电功率的随机波动特性,将符号时间序列方法应用于风电功率波动与预测分析中,并提出一种自适应分区方法,该方法根据数据序列分布的密集程度,实现数据序列区域的非均匀分割,找出信息量丰富的区域,以便突出反映数据的变化情况。之后,以符号序列直方图理论为基础,通过直方图求逆实现原始数据序列关键数据区域的定位,进而完成风电功率的预测。以某一风电场实测风电功率数据验证所提方法的有效性,为风电功率调度提供参考。

Wind power fluctuation analysis and forecasting are one of the major research topics in wind power integration. Considering the randomness of wind power fluctuation, the symbolized time series theory is applied in forecasting analysis, and an adaptive partition method is proposed to realize the non-uniform segmentation of data sequence according to its distribution intensity, by which a region containing more information can be discovered to show the variation of the data. Furthermore, the location of the key data in the original time series can be caught by inversion processing of symbol sequence histograms, leading to the results of wind power prediction. The validity of the method is verified on a wind farm, which can provide reference for wind power dispatching.