基于DBSCAN二次聚类的配电网负荷缺失数据修补  

Repair of missing load data in distribution network based on DBSCAN secondary clustering

作　　者：蔡文斌 程晓磊 王鹏 王渊 CAI Wenbin;CHENG Xiaolei;WANG Peng;WANG Yuan(Inner Mongolia Electric Power Institute of Economics and Technology,Hohhot 010090)

机构地区：[1]内蒙古电力经济研究院,呼和浩特010090

出　　处：《电气技术》2021年第12期27-33,共7页Electrical Engineering

摘　　要：电力负荷属于具有时间序列特性的数据,依据数据固有的规律性和波动性特征,修补由于各种因素而缺失的负荷数据,可为电力系统研究和实验结果的有效性和可预测性奠定基础。本文首先提出基于密度的含噪声应用空间聚类(DBSCAN)二次聚类的方法;其次,提出针对配电网负荷数据的负荷属性相似度,在此基础上进一步提出负荷记录综合相似度;然后,依据DBSCAN二次聚类方法的负荷类别结果和所得负荷记录综合相似度,匹配相似度最大的数据类别,并依据该类别的记录信息对所缺失数据进行修补;最后,采用算例分析证明所提方法的有效性和正确性。Distribution power load belongs to data with time series characteristics.According to the inherent regularity and fluctuation characteristics of the data,repairing the missing load data due to various factors can lay a foundation for the validity and predictability of the power system research and experimental results.Firstly,this paper proposes density-based spatial clustering of applications with noise(DBSCAN)secondary clustering method.Secondly,the load attribute similarity for distribution network load data is proposed,and the load record comprehensive similarity is further proposed.Thirdly,according to the load category results of DBSCAN secondary clustering method and the comprehensive similarity of the obtained load records,the data category with the largest similarity is matched,and the missing data is repaired.At last,the validity and correctness of the proposed method are proved by a numerical example.

关 键 词：基于密度的含噪声应用空间聚类(DBSCAN) 电力负荷 数据相似度 数据修补 

分 类 号：TM714[电气工程—电力系统及自动化]