Abstract
In the age of energy crisis, the scientific and industrial worlds come together to provide solutions and suggestions to reduce energy consumption and respective economic burden. In the energy efficiency direction, problems like non-intrusive load monitoring are still under intense investigation on how to be applied to real-world circumstances with undoubtable results. The current work proposes a framework to overcome existing obstacles in terms of IoT devices involved and algorithmic approaches. To be more precise, the utilization of separate appliance-oriented models is proposed, exploiting heuristic rules that emerge from the very electrical characteristics of the appliances. A hybrid solution is implemented and tested in-vivo for the case of fridges that operate within households, producing promising results and utilizing a powerful yet light smart meter.