Classification of Behavior Profiles for Non-Residential Customers Considering the Variable of Electrical Energy Consumption: Case Study—SAESA Group S.A. Company

Abstract

This work allows characterizing and classifying the consumption profiles of non-residential customers (without distributed generation) based on the consumption curves obtained from the records reported by 934 smart meters in the period from January to December 2019, and which belong to an electric power distribution company in Chile, SAESA Group S.A. To achieve the characterization and classification of the consumption profiles, three typical days are analyzed and determined, which correspond to working days (Monday to Friday), Saturdays, and Sundays or holidays. These three typical days are analyzed for each trimester of 2019. The data processing is carried out on the Power Bi and Matlab® platforms. In Power Bi, the data provided by the electricity company are worked, obtaining the average consumption curves for each client in each period of study considered, while in Matlab®, the visualization and classification of the curves is carried out using the K-means algorithm, to finally obtain the results and conclusions. The results show the existence of seven typical profiles representative of the behavior of non-residential clients, which, in some cases, show similar behaviors, despite being from different categories.

Keywords

• clustering
• non-residential client
• K-means
• load profile
• smart meter