Event Characterization Using Synchrophasor Big Data

Synchrophasor systems have been installed in the US power grid in large numbers over the last 10-15 years and the amount of recorded data is growing exponentially as the PMUs, as well as protective relays, digital fault recorders and other PMU-enabled devices are being added. The amount of recoded data may exceed hundreds of terabytes for a single utility, and is growing every day. Handling of such large amounts of data requires careful considerations of the possible uses of such data, which leads to the requirements for automated means of processing synchrophasor big data files. This panel addresses the task of implementing machine learning and artificial intelligence methods on synchrophasor data for the purpose of developing data models for automatically characterizing power system events. The panelists have first-hand experience with such developments from extensive involvements with the projects aimed at collecting, managing and processing synchrophasor big data. They will share their experiences with the goal of informing the professional communities about the benefits and challenges of implementing the big data analytics. Suh experiences are pointing to many pitfalls of existing practices and suggesting improvements in future synchrophasor big data wrangling, processing and visualization.