基于互信息的电力负荷组合预测模型

Combination model of forecasting power load based on mutual information

针对电力负荷组合预测中权重难以分配问题,运用信息论与模糊评价方法,提出一种新的确定权重的方法。用互信息度量各个预测模型的精度,对所有预测模型的互信息进行模糊评价,并赋予各预测模型相应的权重,互信息大的模型获得较大的权重;反之,获得较小的权重,进而建立一个新的组合预测模型。实例应用表明,经多个误差分析指标的评价,所提出的组合预测模型不仅优于其中任何单个预测模型,而且优于基于熵的组合预测模型。

No good solution to determining the weight coefficient in combination forecasting of power load has been found as yet. This paper attempted to propose a new method for solving the weight coetficient based on information theory and fuzzy evaluation. The accuracy of each forecast model is measured by mutual information. After fuzzy evaluation of the mutual information, each prediction model is assigned an appropriate weight coefficient on the principle that the bigger the mutual information, the bigger the weight coefficient. Numerical experiment shows that the proposed combination forecast model outperforms any of the models in the forecast model and the combination forecast model based on entropy.