Over the past year, I’ve had a successful collaboration with some great colleagues at the Applied Machine Learning Lab at Duke university.
We’ve been investigating finding solar panels from satellite imagery such as pictures from Google Earth. In addition to finding the solar panels, we’ve also been working on methods to estimate their power capacity.
What’s the point? As more individual homes add solar panels, the management
power grids is becoming increasingly complex for power companies,
since the generation of power is no longer entirely under their control.
With accurate estimates of the number of clients who have solar panels and the
amount of power they can potentially feed back to the grid, power companies will be
better able to predict demand and better manage their resources.
Here are the papers we’ve published, both of them authored by Cooper Union students and accepted at peer-reviewed conferences