Automatic detection of solar installation malfunction
Oct 2021 - April 2022
Design of an algorithm to detect and diagnose problems with solar installations.
Underperformance of solar installations can be due to many reasons. We created a ranking algorithm to find similar installations in the companies own portfolio (20K+ installations), and used yield data on a 15 minute time scale to compare and detect anomalies which could be attributed to solar panel malfunctions. The output of the algorithm is a list of potential problems with their statistical likelihood.
My Contribution
- Design and implementation of the algorithms
- Frontend dashboard visualizations
External Links
Highlights
- Data driven detection of solar installation problems
- Huge time saver for the monitoring team
- Important step towards fully automatic monitoring
Users
- Monitoring department of Wocozon
Technology Stack
algorithmsJavascriptPHP/Laravel