Sparse PMU Placement Algorithm for Enhanced Detection and Identification of Power Grid Events

This paper is concerned about the challenge of timely and accurate detection and identification of events involving line outages and faults in power grids. Given the increasing number of installed phasor measurement units (PMU), their measurements are prime sources to be exploited to address this challenge since their high refresh rates allows to capture post- and pre- event phasor values. The challenge is mainly due to the fact that the number of installed PMUs is typically not sufficient to receive measurements from terminal bus(es) of every line. This challenge be addressed by applying model-based sparse estimation methods or data-driven approaches such as convolutional neural networks (CNN) and graphical neural networks (GCN). It should however be noted that success rates of these methods are very strongly related to strategic placement of PMUs, as their locations significantly influence the accuracy of line outage detection and fault identification as noted in <xref ref-type="bibr" rid="ref21" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">[21]</xref>. This dependency constitutes the main motivation of the paper whose focus is development of a strategic PMU placement algorithm to maximize the performance of the event detection and identification methods that use PMU measurements as their inputs. An algorithm that is designed to yield the best possible locations to install a limited number of PMUs, optimizing their utility for model-based sparse estimation methods as well as advanced data-driven approaches is developed and implemented. Performance of the method is demonstrated with only 83 PMUs for the 1354 Bus system, and with 50 PMUs for the 196 bus system.