Energy Disaggregation using Smartmeters
Energy disaggregation, also known as non-intrusive load monitoring (NILM) or appliance-level
energy monitoring is a process that involves breaking down the total energy consumption of a
household into individual appliance or device-level usage. This technique aims to provide a detailed
understanding of how energy is utilized within a home, offering insights into the consumption
patterns of specific appliances such as refrigerators, air conditioners, lighting, and others.
Problem Statement
Energy Disaggregation is an ongoing research topic where the rollout of smartmeters in every
household in Austria offers new challenges. On one hand, smartmeters offer the possibility to use
consumer-friendly “click-on” sensors, on the other hand, time resolution and measurable quantities
differ a lot between devices why the algorithms have to come up for worst case scenarios.
Goal
This master thesis aims to address the challenge of disaggregating energy consumption in
predefined single components yet to be defined (fridge, oven, AC, etc.) in private homes using smart
meter data. The goal is to develop advanced algorithms for the accurate identification and
quantification of individual appliance energy consumption. Overcoming this challenge is crucial for
optimizing energy usage.
Tasks
- Data collection
- Open Source Data is available
- The strategy for collecting household data is already at place and ongoing until the end of the thesis
- Implement a simulation to generate synthetic data
- Sensor
- Compare different “click-on” Sensors and identify the most promising
- Model Creation
- Implement a model that predicts single components from aggregated consumption data
- Optimize and Evaluate Model
- Deployment (optional)
- Deploy to cloud endpoint to get predictions via API request
Requirements
- Familiar with Python and TensorFlow or PyTorch
Offer
- Sponsored Master Thesis with a temporary work contract
- Extension to permanent contract possible
- Extension to PHD Topic with focus on industrial applications possible
- Support and Supervision from an experienced and motivated Team
Your contact person
Viktoria Hummel
Executive Assistant
Email: office@efficientio.com
Phone: +43 664 127 20 74
EfficientIO GmbH
Billrothstraße 4, RG Top 4 & 5
A-1190 Vienna
