A Home Energy Management System with Focus on Energy Optimization
Authors
Tewes, Mikkel Skovsmose ; Foldager, Nicolai ; Hansen, Hans
Term
4. term
Publication year
2018
Abstract
Dette projekt udvikler et hjem-energistyringssystem, der anvender maskinlæring til at reducere el- og varmeforbrug ved at tilpasse styringen af smart plugs, smart lys og en smart termostat til brugerens tilstedeværelse (hjemme, ude eller sover). Prototypen omfatter en hub, en Android-applikation og en webserver som kommunikationslag, og Azure Machine Learning bruges som cloudmiljø til modellering. Systemet indsamler data om strømforbrug og termostatens setpunkter og registrerer tilstedeværelse via telefonens Wi-Fi. To modeller blev trænet til automatisk at slå enheder til eller fra baseret på den afledte status. Resultaterne indikerer et potentiale for energireduktion, men der er behov for mere data og længerevarende tests, før systemet kan afprøves fuldt ud med autonom styring.
This project develops a home energy management system that uses machine learning to reduce electricity and heating by adapting control of smart plugs, smart lights, and a smart thermostat to the user's presence (home, away, or asleep). The prototype includes a hub, an Android application, and a web server as the communication layer, with Azure Machine Learning as the cloud environment for modeling. The system gathers data on power consumption and thermostat setpoints and detects presence via the phone's Wi-Fi. Two models were trained to automatically switch devices on or off based on the inferred status. Results indicate potential for energy reduction, but more data and longer trials are needed before the system can be fully tested with autonomous control.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents