Abstract

Energy efficiency is part of the goals set by governments around the world to reduce the energy footprint and provide sustainable development for all. The arrival of new technologies that allow the monitoring and self-control of electricity consumption within homes, such as smart meters, allow end users to integrate into the intelligent management systems of the electricity grid, by providing information on the flow of energy and prices. This work performs a classification of residential customers from the consumption data obtained from smart meters. For this, a methodology based on the simple K-means algorithm is used to identify patterns of behavior in the consumption of 1179 homes connected to a real low-voltage electricity distribution network in southern Chile equipped with smart meters, and the validation and refinement of the results using numerical measures belonging to the rough sets theory. Final groups are characterized based on their centroids, making possible to convert the large volumes of data collected into useful knowledge, which is beneficial for both the residential customer and the electric power distribution company. The results show that two clusters are the ones that best represent the set of clients. The opportunity to collect consumption data in real time through these devices offers perspectives to optimize public and private policies on electricity distribution.