Building Energy Analytics with Appliance Disaggregation
Work out which appliances are running from a single whole-building electricity meter, then use that breakdown to find waste. Non-intrusive load monitoring is a well defined problem with public datasets and a hard, interesting core.
Advanced
Needs a team, real planning, and a supervisor who knows the area.1 semester, 2 to 3 students
5 to ship
3 optional extrasSuggested stack
What you should ship
- Ingestion and cleaning of high frequency meter data from a public dataset
- Event detection identifying appliance switching from step changes in the aggregate signal
- Disaggregation model attributing consumption to at least 5 appliance categories, evaluated per appliance
- Waste detection identifying always on load and equipment left running outside occupied hours
- Dashboard showing the breakdown, estimated cost per category and identified waste with a value attached
If you have time left
- Detecting appliance degradation from a changing consumption signature over time
- Transfer to a building not present in the training data, with the accuracy drop reported
- Carbon estimation using grid intensity data by time of day
The problem
A building's electricity bill is one number. Reducing it requires knowing where the energy went, and installing a meter on every circuit is expensive. Non-intrusive load monitoring infers the breakdown from the single existing meter.
What you build
Event detection from the aggregate signal, a disaggregation model, and a dashboard that turns the breakdown into identified waste with a monetary value.
Why it is genuinely hard
Many appliances have similar signatures, several run simultaneously, and variable loads such as anything with a motor or a heating element do not produce clean steps. Accuracy varies enormously by appliance, and reporting per appliance rather than as one average number is the honest presentation.
The result that makes it useful rather than academic
Always on load. Most buildings have a constant baseline draw running through the night that nobody has ever quantified, and putting a number and an annual cost on it is often the single most actionable finding. Make that a headline feature.
The transfer question
A model trained on one building usually degrades badly on another, because appliances differ. Testing on a held out building and reporting that drop honestly is more valuable than a strong number on a single building.
Scope warning
Public datasets exist and are the right starting point. Instrumenting a real building is a stretch goal, not the plan.
Ideas and guidance, not finished projects
These are project ideas and scoping guidance, published free for students to use as a starting point. I do not build, write, or sell final-year projects, and I do not complete coursework for anyone. Take an idea, make it yours, and build it.
Not sure this one fits you?
I help students pick an idea that matches their skills and their deadline, then scope it down until it is finishable. Guidance only, never done for you.
Stuck on something specific?
Writing only gets you so far. If you want an answer to your situation rather than the general case, book a session and we will work through it together. Every session is free; a few slots open each week.
Follow along
New writing, resources and project ideas land here first.