基于经验模态分解法的短期负荷分层预测

A Stratified Short-term Load Forecasting Method Based on Empirical Mode Decomposition

电力系统短期负荷预测是电力系统经济调度中的重要内容。目前短期负荷预测的预测方法大多是从总负荷入手进行分析建模,未能充分关注负荷变化的各个组成成分。本文提出1种短期负荷分层预测法。首先,运用经验模态分解法自适应地将总负荷分解为若干个独立的内在模式;然后,对高频、低频和趋势分量分别采用正常日点对点倍比模型、多元回归模型和GM(1,1)模型进行预测;最后,将3个分量的预测结果叠加作为最终的预测值。实例研究表明,该方法有效地提高了短期负荷预测的准确度。

Short-term load forecasting is an important part of economic dispatch in power system. Modeling analysis of the short-term load forecasting method focuses on presently the total load, without paying close attention to each composition. This paper puts forward a stratified short-term load forecasting method. First, decompose the total load into several independent internal model adaptively utilizing empirical mode decomposition （EMD）. Then, predict the high frequency, low frequency and trends component using point to point model, multiple regression model and GM（1,1） model separately. Finally, stack the three components as the final prediction. Case study shows that this method is effective to improve the short-term load forecasting accuracy.