 |
Center for Nonlinear Studies |
Power distribution grids form the final tier in the transfer of electricity from generators to the end consumers. In recent years, distribution grids have become one of the major avenues for smart grid research. This is in part due to influx of new controllable devices, residential generator/storage devices, rooftop solar in residential grids that can potentially be used for cost-efficient operation in the face of uncertainty. Novel dynamic operations in distribution grids is being motivated with the help of smart sensors and IoT (internet of things) based devices with real-time metering capability. However such meters are still not ubiquitous. In this talk I will discuss learning and control problems in distribution grid with smart meter data. Few of the learning methods are generalizable for estimation in related networks like smart buildings. I will then discuss a Markov Decision Process (MDP) framework for network aware control of smart devices in distribution grids that co-optimizes user comfort and global objectives. Finally I will briefly mention a few extensions to problems in power grid research due to increased presence of renewables and smart devices in the grid.
Host: David Metivier