Visualisation and Analysis of
Domestic Electrical Energy Consumption

Imperial Computing Science MSc Group Project Proposal, November 2012

The challenge

As energy costs increase, there is increasing pressure to use energy as efficiently as possible. The first step towards reducing energy consumption is often to measure your existing consumption. To this end, millions of people have installed home energy monitors like this one from Current Cost:

Furthermore, by the end of the decade every home in the UK will have a "smart meter" which will digitally measure energy consumption.

The aim of this project is to help solve two challenges:

One challenge is that data from energy meters is not trivial to interpret, hence energy data is rarely exploited to its full potential. For example, if you were given the graph below could you decide what action to take in order to reduce energy consumption? I know I couldn't!

A second challenge is that there is very little open-access power meter data available to help the research community build better visualisation tools and disaggregation algorithms (disaggregation automatically estimates individual appliance energy consumption from a whole-house meter signal). The research community is hungry for more data.

Two aims of this project

1) Data Visualisation and Analysis Tool

One aim of this project is to produce an attractive, easy-to-use and powerful data visualisation tool to help users understand their own energy consumption data. The raw data might be from whole-house meters, individual appliance monitors or appliance load estimates generated by disaggregation algorithms.

The technologies used to implement these features is your decision but you may like to consider building a web app using modern web technologies like HTML5, SVG, JavaScript and frameworks like d3.js. Web apps have the significant advantage that they can be accessed from a wide range of devices (Windows PCs, Macs, Linux boxes, phones, smart TVs, tablets etc), hence maximising the potential target audience. If your project is successful then we can pay for a web server to take your project "live" (which would be a great project to link to from your CVs).

Exactly which features you implement will be up to you. Some candidate features for your consideration are listed below (and these really are just suggestions. You can implement whatever you want; and you certainly wouldn't be expected to implement all of these features!). Some of the feature suggestions below include questions for research:

2) Record a dataset of power usage

Intel have kindly provided funding to buy each group member enough wireless individual appliance monitors (IAMs) to measure every appliance in your home, and a home energy monitor to measure your whole-house power usage. We have built an open-source wireless IAM base station (based on an Arduino clone called a Nanode - pictured below) and logging software which runs on a laptop or Raspberry Pi.

This dataset will provide a rich set of test data for your data visualisation tool, and will also provide much-needed raw data for the community's research into smart meter disaggregation. And, if you use your own project on a daily basis then you'll inevitably generate lots of great ideas for features, as well as learn lots about the energy consumption of your appliances. Of course, if any group member does not want data from their home to be recorded then this aspect of the project is by no means obligatory!

Users

There is great potential to create a product which will not only gain you a great MSc mark but might also be used by lots of people!

Potential users for the visualisation tool include:

Deciding which users to target will be your choice. If you plan to target a range of users then figuring out how to alter the user experience to match the user's level of geekyness will be an interesting question. Perhaps you could have an "advanced" mode? Or allow users to drill down to more detail if they desire? Or build visualisations which are so intuitive that even novice users can quickly understand the most complex information your system can present (and hence you don't need to worry about separate "geekness" settings)?

There is some existing research looking at how to convey energy data to users. For example, some research suggests that "smiley / surprised faces" are a good way to express "low / high" energy use (e.g.). It might be fun to do some of your own research into this. Perhaps advanced users find smiley faces patronising and annoying whilst novices find them genuinely useful? Or does everyone find them useful?

Both the data visualisation tool and the dataset will hopefully be warmly received by the community. It may be fun to present your data visualisation tool at the monthly Cosm Internet of Things London Meetup (which is very well attended and beer-fuelled) and there may even be scope to produce a workshop paper from your dataset. If you're eager to work with industry then by all means consider contacting an organisation like Cosm, Carbon Culture, open.sen.se or Open Energy Monitor etc. to see if they'd be interested in working with you (and we'll do our best to put you in contact).

Software engineering practice

There is plenty of scope in this project to get involved with a wide range of software engineering techniques, should you wish. For example:

Further reading and existing visualisation tools