A versatile clustering method for electricity consumption pattern analysis in households

Hideitsu Hino, Haoyang Shen, Noboru Murata, Shinji Wakao, Yasuhiro Hayashi

Research output: Contribution to journalArticle

36 Citations (Scopus)

Abstract

Analysis and modeling of electric energy demand is indispensable for power planning, operation, facility investment, and urban planning. Because of recent development of renewable energy generation systems oriented for households, there is also a great demand for analysing the electricity usage and optimizing the way to install electricity generation systems for each household. In this study, employing statistical techniques, a method to model daily consumption patterns in households and a method to extract a small number of their typical patterns are presented. The electricity consumption patterns in a household is modeled by a mixture of Gaussian distributions. Then, using the symmetrized generalized Kullback-Leibler divergence as a distance measure of the distributions, typical patterns of the consumption are extracted by means of hierarchical clustering. The statistical modeling of daily consumption patterns allows us to capture essential similarities of the patterns. By experiments using a large-scale dataset including about 500 houses' consumption records in a suburban area in Japan, it is shown that the proposed method is able to extract typical consumption patterns.

Original languageEnglish
Article number6484217
Pages (from-to)1048-1057
Number of pages10
JournalIEEE Transactions on Smart Grid
Volume4
Issue number2
DOIs
Publication statusPublished - 2013 Mar 26

Keywords

  • Electricity consumption pattern
  • Gaussian mixture model
  • KL-divergence
  • gap statistics
  • hierarchical clustering

ASJC Scopus subject areas

  • Computer Science(all)

Fingerprint Dive into the research topics of 'A versatile clustering method for electricity consumption pattern analysis in households'. Together they form a unique fingerprint.

  • Cite this