基于负荷分解与聚类融合的短期用电负荷预测研究  

Research on short⁃term electricity load forecasting based on load decomposition and clustering fusion

作　　者：马晓琴 马占海 罗红郊 张华铭 MA Xiaoqin;MA Zhanhai;LUO Hongjiao;ZHANG Huaming(Information and Communication Company,State Grid Qinghai Electric Power Company,Xining 810000,China;Beijing Qingruan Innovation Technology Co.,Ltd.,Beijing 100085 China)

机构地区：[1]国网青海省电力公司信息通信公司,青海西宁810000 [2]北京清软创新科技股份有限公司,北京100085

出　　处：《电子设计工程》2023年第20期191-195,共5页Electronic Design Engineering

基　　金：国网青海省电力公司2021年电力市场营销项目(62281420006M)。

摘　　要：传统负荷预测算法通常仅以单一的数据源为基础进行计算,因此在面对动态随机特性较强的场景时难以准确预测。针对这一问题,提出了一种负荷分解后再聚类融合的短期用电负荷预测算法。该算法根据负荷行为按时间顺序分类及分解客户负载,并进行负荷数据的聚类融合,再基于贝叶斯时空高斯过程模型描述不同用电区域间的相关性。同时利用深度学习负荷数据中存在的时空相关性来表征电力消费行为特征,从而实现短期用电负荷的精准预测。在对公开数据集进行的预测实验结果表明,与现有方法相比,所提算法的误差较低,且预测性能显著提高。Traditional load forecasting algorithms usually only calculate based on a single data source,which is difficult to predict accurately in the face of scenes with strong dynamic and random characteristics.To solve this problem,a short⁃term power load forecasting algorithm based on load decomposition and clustering fusion is proposed in this paper.The algorithm classifies and decomposes customer loads in chronological order according to load behavior,clusters and fuses load data,and then describes the correlation between different power consumption areas based on Bayesian spatio⁃temporal Gaussian process model.Through in⁃depth study of the temporal and spatial correlation in the load data to describe the characteristics of power consumption behavior,so as to realize the accurate prediction of short⁃term power load.The experimental results on public data sets show that compared with the existing methods,the error of the proposed algorithm is lower and the prediction performance is significantly improved.

关 键 词：负荷分解 聚类融合 负荷预测 贝叶斯算法 

分 类 号：TP318[自动化与计算机技术—计算机软件与理论] TN998[自动化与计算机技术—计算机科学与技术]