Skip to content

Jiafan Yu

In 2007, the documentary, An Inconvenient Truth made me aware of the challenge of climate change and I began to dream of contributing to a more sustainable society. I had already been working on the development of higher efficiency and lower cost solar cells.

My research focuses on integrating state-of-the-art artificial intelligence tools. Solar panels now account for over 10% of the total electricity generation in some U.S. states, such as
California. But policymakers, utility companies, and engineers still find it difficult to put an accurate number on the country’s total solar power installation, let alone to describe what factors make solar power thrive in certain areas and not others. I am developing a scalable deep learning tool, DeepSolar, that scans high definition satellite imagery across the U.S. to obtain accurate data on distributed solar panels’ location and size.

Analyzing over a billion high-resolution satellite images we found 1.47 million solar rooftop installations in the United States, a much higher figure than previously estimated. By
combining this data set with socioeconomic data, we give unprecedented insight into the societal trends that drive solar power adoption.

DeepSolar is now a publicly available database for energy stakeholders to further uncover solar deployment patterns, build comprehensive economic and behavioral models, and ultimately support the adoption and management of solar electricity. In the future, I plan to create a better mechanism to provide more liquidity to the electricity market and connect the existing wholesale market and the emerging peer-to-peer retail market.

DeepSolar: A Machine Learning Framework to Efficiently Construct a Solar Deployment Database in the United States