A Wider Lens on Energy: Adapting Deep Learning Techniques to Inform Energy Access Decisions (2019-2020)
Critical information for energy access decision-making and electricity system planning is not universally available. Decision-makers require information to determine the optimal strategies for deploying energy resources as well as to decide where to prioritize development and whether electrification should be accomplished through grid expansion, microgrids or distributed generation.
Previous studies have demonstrated that deep learning techniques may be successfully applied to satellite imagery to fill these knowledge gaps. However, these studies have focused on one type of infrastructure or system characteristic and have only been applied to one specific geographic region. They also often require additional training data to adapt to new contexts or identify rare objects.
This Bass Connections project team will work toward filling these critical energy system knowledge gaps to inform the energy access planning and decision-making process.
Using satellite imagery and remote sensing data obtained through deep machine learning techniques, this project team will work to enable the automatic, global collection of data on multiple types of energy infrastructure and electricity access measures. The team will also work to facilitate the fluid transfer of information between geographic regions to increase the potential scale of application, the generalizability of these methods and the ability to accurately identify rare types of infrastructure.
The overarching goals will be to:
- Investigate a new form of domain adaptation by using synthetically generated imagery
- Investigate multiclass energy object identification at scale, including rare objects
- Make data for decision-makers and policy-makers in the energy access and system planning space open source and available for two cities.
New multiclass dataset of energy system infrastructure from multiple countries; tool for the synthetic generation of overhead imagery shared on GitHub; poster and final presentation open to the public; manuscript for publication
Fall 2019 – Spring 2020
- Fall 2019: Begin team meetings and short course on data analytics, energy systems, energy access, research techniques and project management; define project goals and student roles; participate in web-based Kaggle machine learning competition
- Spring 2020: Continue team meetings; engage in skill-based tutorials; engage in informational sessions with guest speakers from academia and industry; develop posters, presentations and manuscripts; disseminate research findings to various audiences
See related Data+ summer project, A Wider Lens on Energy: Adapting Deep Learning Techniques to Inform Energy Access Decisions (2019). See earlier related team, Energy Data Analytics Lab: Energy Infrastructure Map of the World through Satellite Data (2018-2019), and a related Data+ summer project, Energy Infrastructure Map of the World (2018).
Image: Box Springs Wind Farm securing financing through a public private partnership, by Green Energy Futures - David Dodge, licensed under CC BY-SA 2.0
- Kyle Bradbury, Energy Initiative
- Leslie Collins, Pratt School of Engineering-Electrical & Computer Engineering
- Jordan Malof, Pratt School of Engineering-Electrical & Computer Engineering
/graduate Team Members
Vivek Sahukar, Interdisciplinary Data Science - Masters
/undergraduate Team Members
Ayooluwa Balogun, Mechanical Engineering (BSE)
Scott Heng On, Computer Science (BS), Statistical Science (BS2)
Gaurav Uppal, Mechanical Engineering (BSE)
Jason Wang, Computer Science (BS)
/yfaculty/staff Team Members
T. Robert Fetter, Nicholas Institute for Environmental Policy Solutions
Marc Jeuland, Sanford School of Public Policy
Luana Lima, Nicholas School of the Environment-Environmental Sciences and Policy
Robyn Meeks, Sanford School of Public Policy