Smart Home Energy Efficiency
Application of machine learning on smart meter power data to improve efficient electric energy usage in households
M.M.F. Boers (TU Delft - Electrical Engineering, Mathematics and Computer Science)
Kornelis Blok (TU Delft - Energy and Industry)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Nonintrusive Load Monitoring (NILM) can be used to disaggregate household energy
usage collected from a central meter to the level of individual appliances, but has so far mostly been applied in controlled, small-scale settings. Further potential such as the application for energy efficiency classification and management have remained largely untapped. In this research a machine learning model was developed and deployed to determine energy consumption, usage pattern and energy efficiency characteristics of real-life dishwasher usage based on smart meter data. The developed NILM system was deployed on a full year of smart meter data for nearly 130.000 households in the Netherlands. The analysis was accompanied by a survey to gain additional information on the households usage behaviour. The average energy consumption for all households was found to be 1.18kWh per wash, with dishwashers used 240 times per year on average. Dependencies are shown for household size, washing temperature, machine efficiency label as well as time of day, week and year. It was estimated that 9 in 10 households could reduce their dishwasher energy consumption, with an average saving potential of more than 30% per year. The developed method showed to be suitable to gain insight into average electricity consumption and usage patterns on an appliance level, non-intrusively and at large-scale. Additional survey data was shown to provide comparative insight between different user groups. The developed framework can be easily expanded for other major appliances and could be used to drive tailored consumer feedback on energy efficiency improvements within households.