Computer Science Location: Zoom (contact presenter for link)
Add to Calendar 2022-11-28T11:00:00 2022-11-28T11:00:00 America/New_York Ph.D. Defense Presentation by Swapna Thorve Data-Driven Scalable AI for Addressing Problems in the Study of Smart Grids   Abstract: Zoom (contact presenter for link)

Data-Driven Scalable AI for Addressing Problems in the Study of Smart Grids



The wave of grid modernization and climate change is rapidly changing the landscape of residential energy demands. For example, hotter summers imply increased use of A/C units, the use of electric vehicles leads to increased household energy demands, and the use of rooftop solar supports local generation. A central question thus is to understand how energy is consumed at granular social, spatial, and temporal resolutions. Such an understanding can lead to better solutions to demand-response events, study the diffusion process of EV/PV adoption, predict household-level energy use, and/or analyze the impacts of extreme weather. In order to answer these social impact questions, several 'Modeling & Simulation' solutions are appearing in the literature at a noteworthy rate. However, we observe some critical problems that still need to be addressed, especially in the areas of data availability and analyzing complex simulations. Due to these drawbacks, many public policy and social impact questions requiring detailed knowledge of the domain remain unexplored. To facilitate tailored energy policy recommendations and solve problems in fair and sustainable energy, I address these research gaps in my dissertation. The contributions of my dissertation are summarized as follows: (1) Generate a novel large-scale high-resolution digital twin residential disaggregated energy use time series dataset for U.S. households (~30TB); (2) Develop a scalable and extensible big-data pipeline infrastructure using a microservices-oriented architecture for designing a synthetic energy demand modeling simulation; (3) Designing robust validation strategies using machine learning and statistical techniques for evaluating synthetic energy use datasets at multiple resolutions while considering domain knowledge; (4) Solving fairness and sustainability questions in residential energy through agent-based simulations and active learning. Two specific scenarios are considered: (a) diffusion of solar adoption in rural and urban areas; (b) fairness in residential dynamic pricing.



  • Anil Vullikanti, Committee Chair, (CS, Biocomplexity/SEAS/UVA) 
  • Madhav Marathe, Advisor, (CS, Biocomplexity /SEAS/UVA) 
  • Samarth Swarup, Co-Advisor, (Biocomplexity/SEAS/UVA) 
  • Henning Mortveit, (ESE, Biocomplexity/SEAS/UVA) 
  • Madhur Behl, (CS/SEAS/UVA) 
  • Sonia Yeh, (Chalmers University)