Forecasted power load on map
Forecast vs. actual
Service source code on Flowhub
Team Electrocute in action
1.5°C is the global warming target set by the Paris Agreement. How will this affect energy consumption? What kind of generator assets should utilities deploy to meet these targets? When and how much renevable energy can be utilized?
The changing climate poses many questions to utilities. With Electrocute's forecasting suite power companies can have accurate answers, on-demand.
What it does
Electrocute creates energy consumption forecasts using weather and calendar data. Compared to historical records in Ireland, the Electrocute forecasts are within 10.6% error margin (30 amps with the Ireland data set) at any given circumstance.
How we built it
The Electrocute platform comes in three parts:
- Forecasting system built with Python and scikit-learn
- Service back-end built with Node.js and communicating with the forecasting system over MsgFlo message queues
- Visualization front-end built with the Leaflet.js mapping library
In addition we built a Intel Genuino based warning light for high-load forecast situations.
Challenges we ran into
Learning curve for deploying message queue based polyglot microservices on Predix.
Accomplishments that we're proud of
Got it all working! This is the first Flowhub IoT deployment on Predix.
Even has end-to-end continuous integration set up.
What we learned
Strong correlations between weather, day of week, time of day and electricity usage, consistent over a long time.
Connecting different languages and systems in a single Predix deployment.
What's next for Electrocute
Validating the model with utilities and improving error margin of forecasts with more learning data. Go-to-market.