Title: Data-Driven Calibration of Electric Power Distribution System Models
Date: September 30, 2020
Bio: Matthew Reno is a Principal Member of Technical Staff in the Electric Power Systems Research Department
at Sandia National Laboratories. His research focuses on distribution system modeling and analysis with Big
Data and high penetrations of PV. Matthew leads several projects that are applying cutting edge machine learning
algorithms to power system problems. He received his Ph.D. in electrical engineering from Georgia Institute of
Technology.
Abstract: Grid-edge sensing devices, including advanced metering infrastructure (AMI) devices, have enabled
the development of a myriad of novel algorithms focused on calibrating distribution system models. Distribution
system analysis tools are often severely limited in their effectiveness by the accuracy of the model details and
parameters of the grid. This presentation will focus on using grid measurements and Big Data to provide more
accurate feeder model phasing information, parameter estimation, better spatial and temporal load models, and to
detect the presence of distributed energy resources (DER). Synthetic data is used to rigorously test algorithms
under known conditions, and utility data is used to test the algorithms on actual U.S. utility distribution system
models with field measurement data from SCADA, AMI, and other sources. We will also discuss strategies for
managing issues found in utility data, such as missing data and measurement noise, as well as incorporating
physical or domain knowledge into algorithms and algorithm development.