基于时间序列的台区配电负荷峰值预测

Peak value forcasting for district distribution load based on time series

负荷预测是台区配电扩容规划的基础工作。根据台区配电负荷日峰值数据表现出的周期性和渐变性,建立基于时间序列的台区配变负荷峰值预测方法,以某省某台区为例对台区配电负荷峰值进行了预测。模型由自相关和移动平均两个部分构成,根据时间序列的自相关系数与电力负荷数据的先验知识分别确定这两个部分的滞后阶数。预测误差率的分析结果显示,本方法的预测结果精度好,准确性高。该方法克服了台区配电负荷影响因素繁杂、差异性大、难以量化的困难,具有很强的实践性,易于推广。

Electric power load forecasting is the basic task of the power distribution expansion planning in the station. According to the periodicity and continuity of the peak load data of station,a prediction method of the peak load of station based on the time series is established in this paper,and a station in some province is selected as an instance to forecast its load. The model consists of two parts,the autocorrelation part and the moving average part. The autocorrelation analysis and the prior knowledge about load data both determine the lag order of the model. The analysis result of prediction error rate shows that the prediction result has good accuracy. The complicated,diverse and hardly quantified influencing factors of the station distribution load are well handled in this method. Thus,the method is practical and can be easily generalized.