基于局部形相似的超短期负荷预测方法

在对电力负荷局部特性分析的基础上,定义了一个描述负荷序列形状相似程度的指标--负荷曲线形系数,提出了一种基于局部形相似的超短期负荷预测方法.该方法将预测时刻前一段负荷序列与各相似日同一时段负荷序列的形系数引入到超短期负荷预测中,强调基于形相似基础进行值预测,克服了现有预测方法中对各相似日采用相同权重所导致的平滑效应对拐点负荷预测的影响.研究结果表明,该方法在保证运算速度的同时,提高了总体预测准确性和拐点处的预测准确性.该改进方法可用于对现有多种超短期负荷预测方法的改进.

Very Short-Term Load Forecasting Method Based on Local Shape Similarity

Based on the analysis of power load local characteristic, load curve coefficient is defined as an index which can describe load series shape similarity. And a method for very short-term load forecasting is presented based on the local shape similarity. In the proposed method, the shape coefficients between the load series ahead of the forecasting time in the forecasting day and the series at the same period of time in the similar days are introduced, which emphasizes on that the value forecasting should be based on the shape similarity. The influence of smoothing effect on the inflexion point load forecasting caused by using homology weights is overcome by the proposed method. Study results show that the proposed method can improve the global forecasting accuracy including the inflexion point while keeping the calculation speed, and it can be used for improving the existing methods of very short-term load forecasting.