多变量数据聚类最优选择的用电关联分析算法
DOI:
作者:
作者单位:

1.国网山东省电力公司济南供电公司;2.清华大学 电机系;3.北京汇思慧能科技有限公司

作者简介:

通讯作者:

中图分类号:

TM711

基金项目:

国网山东省电力公司科技项目(2020A-007)


Electricity Consumption Association Analysis Algorithm for Optimal Selection of Multivariate Data Clustering
Author:
Affiliation:

1.Jinan Power Supply Company,State Grid Shandong Electric Power Company;2.State Key Lab of Control and Simulation of Power Systems and Generation Equipments Deptof Electrical Engineering,Tsinghua University;3.Beijing Huisi Huineng Technology Co,Ltd

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对多源用电大数据典型性分析结果不唯一且精度不高等问题,提出了一种多变量数据聚类最优选择的用电关联分析算法。算法借助小波变换日负荷聚类实现多源用电日负荷的相似性聚类,以提高数据分析的准确性;然后在获得分组上进行单次细粒度典型相关分析;利用典型相关分析的预测性验证典型权重准确性,以实现单次分析结果的最优选择,实现分析结果的唯一性。算法在北京地区非居民用电客户的用电、用气和天气三元数据集上仿真实验,结果发现在不同用户群体上三元数据的典型性相关曲线存在基本稳定、季节性和周期性变化等三种模式。与其他八种算法对比可知,本文算法的关联挖掘最为深入和准确,其中平均相关系数至少提高了1.52%,均方差误差至少降低了2.09%。

    Abstract:

    Aiming at the problems of non-unique and low accuracy of typical analysis results of multi-source electricity consumption big data, The electricity consumption association analysis algorithm for optimal selection of multivariate data clustering is proposed. The daily load clustering of wavelet transform is used to realize the similarity clustering of the daily load of multi-source electricity consumption, thereby improving the accuracy of data analysis;Then,a single fine-grained canonical correlation analysis is carried out on the obtained group, and the accuracy of canonical weight is verified by the predictability of canonical correlation analysis, which obtain the optimal selection of single analysis result and realize the uniqueness of analysis result.The algorithm is simulated on the electricity, gas and weather data sets of non residential electricity customers in Beijing.The results show that there are basically stable, seasonal and periodic changes in the typical correlation curve of the three data sets in different user groups.Compared with the other eight algorithms, it can be seen that the association mining of the algorithm in this paper is the most in-depth and accurate, in which the average correlation coefficient is increased by at least 1.52%, and the mean square error is reduced by at least 2.09%.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2021-04-10
  • 最后修改日期:2021-07-10
  • 录用日期:2021-09-11
  • 在线发布日期:
  • 出版日期: