Energy Hack NL 2018 - Team Granny Heckle

---

About the hackathon

How can we make sure that all of the energy that is used in the Netherlands is 100% sustainable? Custom solutions, data and technology can be a solution here. Enpuls and Enexis Netbeheer cooperate with TU Eindhoven, Microsoft, TNO, FAN, UtrechtInc, 80 hackers and innovators to come with innovative solutions to three challenges: Flexible Energy, Durable Urban Development and Durable Mobility.

More information on Energy Hack NL

---

Our idea: Sonnet

We developed a product called Sonnet, which is a solar panel prediction system that can help prevent and reduce station failure. We have built the following features:

Solar panel database<br> We have built a big database containing solar panel data. We have enriched an existing registration of solar panels by detecting non-registered solar panels in aerial photography (Open Data NL) with a deep neural network.

Demographic database<br> We have also created a database with prognostic demographic data based on existing demographic data on a district level. We predicted the demographics per district for the upcoming 10 years. We also added the average roof quality per district by scraping the ZonAtlas, which tells you whether your roof is suitable for solar panels.

Solar panel forecast<br> We combine the two databases to predict solar panel growth on a district level.

Station capacity forecast<br> Finally, we can combine all these effort into a product that shows if a station is prone to failure by comparing the predicted net capacity used by solar panels in a certain district and the available capacity in the district stations.

---

Meet our team!

Robbert van der Gugten, data science consultant at Big Data Republic <br> Steven Reitsma, data science consultant at Big Data Republic <br> Joris van Vugt, data science student at the Radboud University Nijmegen <br> Tanja Crijns, data science student at the Radboud University Nijmegen <br> Chris Kamphuis, data science student at the Radboud University Nijmegen <br>

---

Overview of data sets used

• Aerial imagery from Open Data NL
• Groningen solar panel data (to train our solar panel detector)
• Amsterdam solar panel data (to train our solar panel detector)
• Demographics data from CBS (2011-2017)
• ZonAtlas
• Maximum capacity of grid stations in Groningen (supplied by Enexis)
• Average load of grid stations in Groningen (Open data Enexis)

---

---

Extended version of our final presentation