基于相似日负荷修正的节假日短期负荷预测

Short-term load forecasting for holidays based on the similar days’ load modification

针对短期负荷预测中节假日的特殊性,提出基于相似日负荷修正的算法。为了克服节假日样本较少所导致的预测困难,将近期周末扩充到相似日样本中。该算法评估相似性时综合考虑气象和时间因素。针对气象因素,采用灰色关联分析法来计算,而对于时间因素,兼顾到'近大远小'和'周期性'的原则,并以二者相似度的乘积量化总体相似性,选出若干相似日。考虑到类型日和年周期对负荷的影响,基于'历年节假日与周末负荷比例'和'负荷年增长率'分别修正相似周末和相似节假日的负荷。该算法应用到某地级市的预测中,平均相对误差为2.29%。

According to the specificity of short-term load forecasting for holidays, an algorithm is proposed based on the similar days＇ load modification. In order to overcome the difficulty due to few samples, recent weekends are added to the samples from which similar days are selected. Moreover, when it comes to assess the similarity, this algorithm takes into account the meteorology and time. The similarity in meteorology is obtained with grey correlation analysis method, while the similarity in time takes into consideration of ＂near greater far smaller＂ and ＂cyclical＂ principle. Furthermore, the product of these two is to evaluate the overall similarity in terms of which the similar days are selected. In addition, considering that the day type and annual cycle have an impact on the load, the load of similar weekends is modified on the basis of the proportion between the previous holiday and corresponding weekend while that of similar holidays is amended on grounds of the rate at which load grows. Finally, this algorithm is put into practical operation, and it offers an average relative error of 2.29%.