I programmed a Raspberry Pi with a machine-learning model that analyzes over 1,500 data points, tracking the sun’s trajectory, seasonal variations, and surrounding shading patterns.
Using real-time data from my Victron system and open-access satellite weather data from Norway, it continuously learns to predict my solar electricity production with over 90% accuracy.
Tomorrow’s forecast: 50 kWh, just 31% of the clear-sky potential.
Amazing what a little Raspberry Pi combined with @Victron_Energy’s open-source ecosystem can do!
Everyday life
A Raspberry Pi learned to forecast an off-grid home's solar output
Dr. Simon, a biotechnologist living off-grid in Brazil, trained a model on a Raspberry Pi with 1,500 data points covering the sun's path, seasons and shading. Fed live Victron data and Norwegian satellite weather, it predicts the next day's solar output with over 90% accuracy.
363likes
13reposts
55saves
16Kviews
- Made with
- Raspberry Pi, Victron
- Role
- Biotechnologist
- Posted
