JUCS - Journal of Universal Computer Science 24(9): 1306-1329, doi: 10.3217/jucs-024-09-1306
The Generation of Electricity Load Profiles Using K-Means Clustering Algorithm
Rūta Užupytė‡, Tomas Babarskis§, Tomas Krilavičius|‡ Vilnius University, Vilnius, Lithuania§ Binar Solutions, UAB, Vilnius, Lithuania| Vytautas Magnus University, Vilnius, Lithuania
Corresponding author: Rūta Užupytė ( ruta.uzupyte@bpti.lt ) © Rūta Užupytė, Tomas Babarskis, Tomas Krilavičius. This article is freely available under the J.UCS Open Content License. Citation:
Užupytė R, Babarskis T, Krilavičius T (2018) The Generation of Electricity Load Profiles Using K-Means Clustering Algorithm. JUCS - Journal of Universal Computer Science 24(9): 1306-1329. https://doi.org/10.3217/jucs-024-09-1306 |  |
Abstract
Accurate information about the actual behavior of electricity users is essential to the electricity suppliers in order to ensure efficient decisions in planning pricing, e.g., designing tariffs and load planning. Load profiles of customers is a straightforward source for such data, however it should be analyzed to extract relevant information. Most of the existing techniques are tested with small data sets or over short periods, which does not allow to investigate seasonality influence. We present a new methodology for the grouping of electricity customers based on the similarities of their (hourly) consumption patterns. Approach is based on the periodicity analysis and well-known clustering technique - K-means, which is applied for identification for separate users load profiles and clustering of load profiles. Values of model parameter are selected using adequacy measures. Finally, the results obtained by this methodology with a data set of 3753 electricity customers are presented, and future plans discussed.
Keywords
electricity patterns, load profiling, time-series clustering, clustering technique