配电网短期负荷分频段组合预测方法

Research on Combined Forecasting Method by Dividing Frequency Band in Short-Term Load of Distribution Network

以小波分析理论为基础,研究分频段组合预测方法。选用db4小波函数将历史负荷序列进行最大尺度分解重构,根据不同频段负荷分量的特点,对低频分量用GM(1,1)预测,中频分量用时间序列法预测,高频分量用二次指数平滑法预测,然后将各频段分量的预测值叠加,得最终预测结果。以福建省某10 kV线路短期供电量及EUNITE竞赛数据为例,通过Matlab仿真程序,对比分析3种单一预测结果和分频段组合预测结果,结果表明,分频段组合预测效果明显优于单一预测,且具普遍性。最后,用某市局配电网短期负荷数据,验证分频段组合预测法是有效可行的。

This paper mainly studies the combination forecasting method by dividing frequency band based on wavelet analysis, which uses the db4 wavelet to decompose and reconstruct the history load date, according to the characteristics of different frequency components, the algorithm uses GM （ 1,1 ） to predict the low frequency com- ponents of the load, uses the time series method to predict the intermediate frequency components of the load, uses the secondary exponential smoothing method to predict the high frequency components of the load. Then, the load forecasting values of each frequency component are summed up to produce the load forecasting result. Taking the short-term power supply of one l0 kV line in FuJian Province and the date of EUNITE as examples, there is a com- parative analysis of three kinds of single forecasting results and the combined forecasting results, showing that the combined forecasting effect is better than the single prediction. Finally, the effectiveness and the feasibility of the combined forecasting method in this paper are proved by the short-term load date of one power supply bureau.