Title: Provable estimation in distribution grids: a physics-informed statistical learning perspective
Date: June 23, 2020
Bio: Deepjyoti Deka is a staff scientist in the Theoretical Division at Los Alamos National Laboratory, where he
was previously a postdoctoral research associate at the Center for Nonlinear Studies. His research interests include
data-analysis of power grid structure, operations and security, and optimization in social and physical networks.
At LANL, Dr. Deka serves as a co-principal investigator for DOE projects on machine learning in power systems
and in cyber-physical security. Before joining the laboratory he received the M.S. and Ph.D. degrees in electrical
engineering from the University of Texas, Austin, TX, USA, in 2011 and 2015, respectively. He completed his
undergraduate degree in electrical engineering from IIT Guwahati, India with an institute silver medal as the best
outgoing student of the department in 2009.
Abstract: Distribution Networks provide the final tier in the transfer of electricity from generators to the end
consumers. In recent years, smart controllable devices, residential generator/storage devices and distribution grid
meters have expanded the availability of sensor data in distribution networks that can be used for different
learning/estimation problems. Such problems include topology identification, line impedance and load statistics
estimation, phase identification and others. For a range of realistic operating conditions, including ones with
partial observability, we develop learning algorithms by merging tools from statistical machine learning with
physical laws related to static and dynamic operation in power grids. Use of statistical methods produces provably
consistent estimation in the large sample regime, but also enables us to give robust guarantees on the performance
in the finite-sample and noisy regimes. Additionally, I will discuss how statistical machine learning can work in
conjunction with newer neural network based methods for estimation in other related cyber-physical networks
such as gas systems, sensors and smart buildings.