摘要
在电网的低压台区,在用户接线改动或者为了负荷均衡分配而进行线路改造后,由于信息更新不及时,经常导致用户进线端与集中器归属关系记录不准确、台户关系与实际不符等问题。单纯依靠人工去识别台户关系不仅费时费力而且效果不佳,同时需要成本高昂的专用设备。为解决上述问题,文章提出一种基于用电大数据的台户关系智能识别方法。首先搭建台区模型,并采集和预处理台区各用户的海量电压时序数据作为观测变量;然后采用FastICA技术对处理后的数据进行独立成分分析与特征提取,从而获得用于估计观测变量的一系列相互独立的随机变量和混合矩阵;最后利用K均值聚类的方法对特征提取后数据进行聚类分析,从而实现台户关系识别。同时文章对电压数据采样精度和数据量对结果的影响进行了分析比较。仿真算例表明,所提方法能够在保证精度的情况下仅采用电压时序数据就实现台户关系的智能识别。
In the low-voltage transformer area,with the change of user wiring or the modification of line due to the balance of load distribution,because the information update is not timely,the record of the attribution relationship between user incoming line and concentrator is inaccurate,and the actual relationship is inconsistent.Manual identification of user relationships in the station is timeconsuming and labor-intensive,and the cost of dedicated equipment is high.In order to solve the above problems,this paper proposes a method for identifying relationship between transformers and users based on electricity big data.Firstly,a low-voltage transformer station model is built and the voltage time series data of each user in the station is collected and processed as an observed variable. Then,using the FastICA technique,the processed data is analyzed by independent component analysis and feature extraction to obtain a series of independent random variables and mixing matrices for estimating the observed variables.The K-means clustering method is used to aggregate the date after feature extraction.Finally,the influence of voltage data sampling accuracy and data volume on the results is analyzed and compared.The case study shows that the proposed method can realize the intelligent identification of the relationship between transformers and users under the condition of ensuring accuracy only using the voltage time series data.
作者
黄旭
王伟恒
吴双
胡伟
HUANG Xu;WANG Weiheng;WU Shuang;HU Wei(State Grid Liaoning Electric Power Company,Shengyang 110004,China;State Key Lab of Power Systems,Dept.of Electrical Engineering,Tsinghua University,Beijing 100084,China)
出处
《供用电》
2019年第10期22-29,共8页
Distribution & Utilization
基金
国家电网有限公司科技项目(SGLNSY00FZJS1900513)~~
关键词
大数据
关系识别
低压台区
独立成分分析
时序数据聚类
big data
relationship recognition
low-voltage transformer areas
independent component analysis
time series data clustering
