We have been working with CarbonCoop (based in Manchester, UK) for a number of years on various retrofit and smart-grid monitoring projects, see our blog posts and CarbonCoop Blog posts. the CarbonCoop are doing amazing work in real world housing retrofit and smart grid research. See CarbonCoop member @zapaman’s twitter account for what this looks like in practice.
In January Carbon Coop have a masters student doing a thesis on demand/generation forecasting using machine learning algorithms. He is going to use electricity usage/generation data to predict flexibility 1-/2-/3-day ahead.
For the research we are looking for 50-200 participants to provide a emoncms.org read API key and an approximate geographical location.
The project will create a virtual community smart grid by using data collected from energy monitors and smart meters to forecast electricity demand and generation in a group of consumers using historical and live electricity and weather data. This is useful for helping to predict times when electricity demand will be matched to local and national renewable generation which is important for creating a more flexible electrical grid and increasing the utilisation of renewable energy.
The project will begin in February 2017 and we are seeking interested volunteers from amongst the Carbon Co-op membership and others to participate. Those involved will need to have an existing Open Energy Monitor or will be provided with one by Carbon Co-op at a discounted price. Results and software developed for the project will be open source and contributed back to projects like Open Energy Monitor and emoncms.
To get involved you can fill out the following simple form. Or send an email to [email protected].
We all have some kinda setup but for sure not a standardised one.
What kinda setup is needed for the study, is it a reading from what is coming in the house from the grid only ? As several off us also generate alternative energy so only using grid consumption is underestimated when we produce/use locally …
Hi thanks for your interest. It just needs a feed of your electricity consumption from the grid at the meter point (such as the one tagged with ‘use’) and a solar generation feed (if you have one).
Hi Dave, the main thing we are interested in with this is the electricity consumption and any generation from PV (or micro wind turbines although it is unlikely we will get any). So ‘Supply’ and ‘Total Generation’ should do the job. To upload these to emoncms.org you need to have an account on emoncms.org and then you need to copy the ‘read & write’ api key from there into your emonhub,conf. This will then automatically start pushing your data to emoncms.org
After that your feeds from your local monitor should appear as inputs in emoncms.org. There is then a separate ‘read’ api key on emoncms.org which is what we would need to read the data.
then I need to add some more as I only monitor solar generation and what is taken from the grid to power the batteries if no solar power is available
Going to have a look at what it would cost to add an emontx and all needed to monitor à 3 phase system in France
Hello everyone,
this is Gonxo the student running the experiment. I will be happy to answer your questions regarding the project so do not hesitate to PM me. I will try to come over here once in a while so even though I will try to answer in this channel maybe is not the fastest one.
Thanks for your intererst in participate.
What is needed are the feeds tagged properly. With the label “use” for the power consumption and “solar” for the panels. If you have an extra source of energy not documented by the smart meter it will be usefull to know.
digging up this old thread.
Since 6 months now you have data from some of us (in any case mine) but I never ever saw anything coming back. Is it possible to give us an update on this case ?
Hi Eric, @GonxoMR completed his work on this last month and submitted what he had done as his masters thesis (perhaps this is available somewhere publicly now?). The code he has produced as part of it is available and we are examining how this could be integrated into emoncms in the future.
As @GonxoMR has started a job and we currently do not have anyone working on this (apart from me when i have bits of time) and it is unlikely much progress will be made over the summer, but I am looking for another student and/or some funding to continue the work in the near future. Also, I know University of Sheffield are looking at this work as part of their own forecasting algorithm and I have sent them the code.
The next step (in my opinion) is to produce a dashboard showing the live aggregated results across ~10 houses along with the predictions. I think this is maybe another 10 hours work for the right person. If anyone else is interested in working with us to develop this further it would be good to hear from you. We need skills/knowledge in two areas; machine learning algorithms and front end web dev.
