Development of electricity consumption profiles of residential buildings based on smart meter data clustering

In the present research, a high-resolution, detailed electric load dataset was assessed, collected by smart meters from nearly a thousand households in Hungary, many of them single-family houses. The objective was to evaluate this database in detail to determine energy consumption profiles from time series of daily and annual electric load. After representativity check of dataset daily and annual energy consumption profiles were developed, applying three different clustering methods (k-means, fuzzy k-means, agglomerative hierarchical) and three different cluster validity indexes (elbow method, silhouette method, Dunn index) in MATLAB environment. The best clustering method for our examination proved to be the k means clustering technique. Analyses were carried out to identify different consumer groups, as well as to clarify the impact of specific parameters such as meter type in the housing unit (e.g. peak, offpeak meter), day of the week (e.g. weekend, weekday), seasonality, geographical location, settlement type and housing type (single-family house, flat, age class of the building). Furthermore, four electric user profile types were proposed, which can be used for building energy demand simulation, summer heat load and winter heating demand calculation. (c) 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Automated summary

This research assessed a high-resolution electric load dataset from nearly a thousand households in Hungary, using different clustering methods and validity indexes to identify energy consumption profiles. The k means clustering technique was found to be the best method, and analyses were conducted to identify different consumer groups and parameters affecting energy consumption.