Residential electrical consumption disaggregation on a single low-cost meter

Abstract

Demand and cost of electricity is expected to grow in the next years. This has raised interest in monitoring energy usage to reduce losses, and to provide real-time feedback about the cost of the electrical power consumed. This paper focuses on the implementation of a stand-alone system capable of real-time tracking of the power used and that provides power consumption estimation for each device from a single point of measurement. The learning activity is done by detecting the possible state of the electrical devices using a clustering algorithm, which involves k-means technique to analyze and detect the state of an appliance.