基于Prophet-LightGBM的台区短期负荷预测方法  

The Short-term Load Forecasting Method of Transformer Based on Prophet-LightGBM

作　　者：徐岩[1] 张晓 周兴华 孙杰 仇向东 谢理强 XU Yan;ZHANG Xiao;ZHOU Xinghua;SUN Jie;QIU Xiangdong;XIE Liqiang(School of Electrical and Electronic Engineering,North China Electric Power University,Baoding 071003,China;Beijing Join Bright Digital Power Technology Co.,Ltd,Beijing 100085,China)

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003 [2]北京中恒博瑞数字电力科技有限公司,北京100085

出　　处：《华北电力大学学报（自然科学版）》2024年第6期13-19,共7页Journal of North China Electric Power University：Natural Science Edition

基　　金：国家重点研发计划(2022YFE0105200).

摘　　要：台区负荷短期预测对于配电网优化运行、精细管理以及节能降损均有着重要的作用。为了在短期内对台区负荷进行准确的预测,提出一种基于Prophet-LightGBM的台区短期负荷预测方法。它利用二次预测思想,综合Prophet和LightGBM两种算法的优点,实现了更精准的预测。首先对台区负荷进行特性分析,分析某台区负荷在一周内的具体变化情况;然后采用Prophet算法进行历史负荷拟合与第一次预测,提取台区负荷中变化较规律的特征项;最后利用上述特征项以及LightGBM算法对台区负荷进行第二次预测,已达到预测台区负荷整体变化的目的。实验结果表明,与Prophet、LightGBM预测的结果相比,提出的方法在三种类型的台区上均具有更高的预测准确性。Short-term load forecasting of transformer plays an important role in optimal operation,fine management,energy saving and loss reduction of distribution network.In order to accurately forecast the transformer load in the short-term,we proposed a short-term load forecasting method of transformer based on Prophet-LightGBM.It used the idea of secondary forecasting and combined the advantages of Prophet and LightGBM to achieve more accurate forecasting.Firstly,we analyzed the characteristic of the transformer load and the specific changes of a transformer load in a week.Then,we used Prophet’s algorithm for historical load fitting and the first prediction to extract the characteristic items of the transformer load with more regular changes.Finally,we used the above-mentioned characteristics and LightGBM algorithm to forecast the transformer load for the second time,so as to achieve the goal of predicting the overall change of transformer load.The experimental results show that compared with the prediction results of Prophet and LightGBM,the proposed method has higher prediction accuracy in three types of transformers.

关 键 词：PROPHET LightGBM 二次预测 短期预测 台区负荷 

分 类 号：TM715[电气工程—电力系统及自动化]