Close

Search

Search Menu
Photonics Media Photonics Buyers' Guide Photonics Spectra BioPhotonics EuroPhotonics Vision Spectra Photonics Showcase Photonics ProdSpec Photonics Handbook
More News

Machine Learning Identifies Nearly All US Solar Panels from 1 Billion Images

Facebook Twitter LinkedIn Email Comments
STANFORD, Calif., Dec. 28, 2018 — Stanford University scientists developed a machine learning program that analyzed more than 1 billion high-resolution satellite images and identified nearly every photovoltaic solar power installation in the contiguous 48 U.S. states. They found 1.47 million installations, more than previous estimates. Their results, which include activators and impediments to solar deployment, are publicly available on the project’s website.

The team provided its machine learning program, named DeepSolar, with about 370,000 images. Each image was labeled as either having or not having a solar panel present. From these images, DeepSolar learned to identify features associated with solar panels — for example, color, texture, and size. “All of these need to be learned by the machine,” said researcher Jiafan Yu. DeepSolar learned to identify images containing solar panels with 93 percent accuracy. 

Machine learning program identifies nearly all solar installations in U.S.
Stanford researchers created an accurate deep learning model for detecting solar panels on satellite imagery and built a nearly complete solar installation database for the contiguous U.S. Courtesy of Stanford University.

The researchers then had DeepSolar analyze 1 billion satellite images to find solar installations. DeepSolar got the job done in 1 month — work that would have taken existing technology years to complete.

The resulting database contains residential and business solar installations, as well as many large, utility-owned solar power plants. DeepSolar was directed to skip over the most sparsely populated areas, where solar panels are not likely to be attached to the grid. Based on the data, the scientists estimate that 5 percent of residential and commercial solar installations exist in these rural areas.

The researchers also identified key environmental and socioeconomic factors correlated with solar deployment. They developed high-accuracy machine learning models to predict solar deployment density, utilizing these factors as input.

The researchers plan to expand the DeepSolar database to include solar installations in rural areas and in other countries with high-resolution satellite images. They also intend to add features to calculate a solar installation’s angle and orientation, which could accurately estimate its power generation. DeepSolar’s measure of size is for now only a proxy for potential output.

The group expects to update the U.S. database annually with new satellite images. The information could ultimately feed into efforts to optimize regional U.S. electricity systems, including the Stanford research team’s project to help utilities visualize and analyze distributed energy resources.

The research was published in Joule (https://doi.org/10.1016/j.joule.2018.11.021).

Photonics.com
Dec 2018
Research & TechnologyeducationStanford UniversityAmericasimagingphotovoltaicsmachine learningsolarenergyenvironmentDeepSolardeep learning

Comments
back to top
Facebook Twitter Instagram LinkedIn YouTube RSS
©2019 Photonics Media, 100 West St., Pittsfield, MA, 01201 USA, info@photonics.com

Photonics Media, Laurin Publishing
x We deliver – right to your inbox. Subscribe FREE to our newsletters.
We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.