Group MSc project on "Visualisation and Analysis of Domestic Electrical Energy Consumption"

I'm hoping to organise a Computer Science MSc group project on "Visualisation and Analysis of Domestic Electrical Energy Consumption". One aim of the project is to produce an open-source app tol help people save energy by producing useful and interesting visualisations (pretty graphs) of their energy usage. The ultimate aim is to produce something a bit like Google Powermeter. But better ;)! The full spec of the group project is available here.

I'd be really eager to hear any comments / suggestions at this (very) early stage to try to make the project both lots of fun for the students and also useful to the community. Some specific questions might be:

  • Do you currently use a tool to visualise your energy use? (Such as Current Cost's website or Cosm or open.sen.se or something similar?)
  • Are there any specific statistics / visualisations that you'd be particularly interested in?
  • Would you be interested in being able to share details of your energy use with select friends to compare progress?

(To give a bit of context: Group projects run for the duration of the spring term (11 weeks starting Jan 2013). I'm a PhD student in the computing department so I'll be supervising the project but the detailed design is up to the MSc students. Groups usually consist of 5-6 MSc students. There are no guarantees that a group will actually want to do the project, of course!)

Comments

Looks like a really cool project Jack, I wish I'd done this for my fourth year project! Although almost any visualisation is interesting to techies like us, I'd be really interested to see which visualisations inspire the greatest reduction in energy consumption. Ideally, I'd say the visualisations should highlight the easiest ways to save the most energy, not just inform the occupants of things they can't change!

Hey Oli! Very good point that visualisations should highlight the easiest ways to save the most energy.

I met with a group of MSc students interested in the project earlier today. Very enthusiastic, bright folk: I hope they pick this project. But guess what: one of their main interests is machine learning so they're interested in disaggregation! I was thinking of suggesting that they try re-implementing George Hart's algorithms as an open-source project, both because that should be a relatively low-risk way to get some level of disaggregation working within the relatively short amount of time that they have available (and these group projects are more about producing a polished finished product rather than doing cutting edge research), and also because it might make an interesting workshop paper. I'd be really interested to hear your thoughts on that: if you were running a group project, where they had only 10 weeks (and they will have relatively little time to spend on their group project because next term is very intensive with other courses), what disaggregation approach would you recommend?

Yeah that sounds like a great idea. George Hart actually proposed many methods (manual/automatic set up, binary/multi-state appliance models, combinatorial optimisation and/or switch continuity principle). These are all good ideas, but implementing and testing them all will take far too long. I'd suggest picking one or two simple approaches and investigate whether you can validate's Hart's hypotheses through empirical testing. It seems likely that you can continue to increase the accuracy of an approach through the addition of more complexity, but it would be interesting to see which extensions give the greatest improvements. Kim et al. (2011) do this sort of comparison very well.

That's an excellent idea to compare different extensions, as per Kim et al 2011. Fingers crossed that we actually get a student group ;)

That sounds like a very interesting project! Personally, I'm not using anything more than RRD to track consumption, which is a step up from just an energy display. Yes, this means that my data stays under my control. Working in infosec means one of my concerns is ensuring that collected data is unlikely to be connectible to a physical real world address, as it would certainly give a good indication of occupancy. That makes it a bit more difficult, although I'd probably be happy to share that information with a few people (IE, those I might trust to tell I'm away on holiday or work trip).

Just a quick update to say that I've been very lucky! A group of 6 excellent MSc students have selected the Energy project. We've had our initial meeting this term but work won't start properly until next term. It's not known yet whether or not they will focus on disaggregation or if they'll focus more on data visualisation (they will decide early next term).

Now I just need to finish building the meters they'll be installing... ;)

Ah great news! I'd be really interested to hear what direction they decide to take the project in. Keep me informed! By the way, the myJoulo project we talked about has launched now. Check it out: https://www.myjoulo.com/ I hope it tells you your insulation was worth it!

Hi Oli,

Sure thing, I'll definitely let you know how it progresses. And thanks loads for letting me know about MyJoulo - I'll sign up tomorrow!

Thanks, Have a good Christmas, Jack

Hi Jack,

Just wondering how this group project went? I'm guessing it must be nearing its conclusion by now. I ask because I might be co-supervising an MSc individual project and it would be good to see what others have achieved given this short time frame.

Cheers,
Oli

Hi Oli,

They are currently very hard at work on it! The deadline for the report is the evening of Fri 15th March. I think the group will be able to share their report (and possibly their code) although I'm not certain.

I'd definitely recommend co-supervising MSc students. It's nice to get a new perspective. But, of course, don't underestimate the amount of your time it can consume. My MSc students have been fantastic and haven't required much input from me but it's taken bloody ages for me to get their power monitoring kit built ;). I'm really looking forward to being able to focus fully on disaggregation again after spending several months building our data collection system.

Hi Oli,

If you want to see an MSc individual project on disaggregation then - if you'll excuse the self promotion - you could take a look at my MSc project ;)

Add new comment