Monitoring of home appliances provides useful information on how to improve user consump-tion habits and refine energy conservation. This information could also be used in combina-tion with other energy conservation software to reduce energy consumption overall. Manyhave tried to measure the power consumption per device using sub-meters at plug level, thishowever is impractical and economically infeasible. In this paper we explore the possibility ofobtaining the state of devices from aggregate power consumption data using neural networks.We will explore an existing method and analyse how well it does on the most frequently presentdevices in a household and extend it to take weather data into account. In addition to thatwe will explore the opportunity of generalizing this implementation across houses by lettingthe neural network learn on a set of household consumption data and testing it on anotherhousehold
the main implementation and tests have been made.