A Transfer Learning Framework for Power System Event Identification

Abstract

The increasing uncertain components of power systems foster the wide applications of Machine Learning (ML) techniques. While traditional ML models demand a large set of data, data-scarce dilemmas exist for new meters, devices, and new grids. Further, for rich historical measurements, valuable data may still be limited, especially for targets like identifying system events that rarely occur in the power system. To enhance the event type differentiation and localization for a datalimited grid, we propose a Transfer Learning (TL) framework to transfer knowledge from a data-rich grid (source grid) to the target grid, using measurements from Phasor Measurement Units (PMUs). The transferring process is challenging because of (1) high-volume data with redundant information, (2) different measurement dimensionalities, (3) dissimilar data distributions, and (4) disjoint event-location-label spaces for two grids. To handle the challenges of (1) to (3), we propose a joint optimization to reduce dimensionality and maximize common knowledge in a shared low-dimensional feature space, where the commonality lies in the same dimensions and close data distributions. Such an optimization-based procedure is verified via rigid mathematical theorems given the same label space, i.e., event-type-label space. However, for event localization, challenge (4) obstructs the optimization. Therefore, we design a label space alignment method to relabel the event location by the event zone location and build an event zone estimation problem. Then, the framework is generalized to both tasks. Finally, comprehensive experiments demonstrate the advantages of the proposed methods over stateof-the-art transfer learning models.

Structural Tensor Learning for Event Identification With Limited Labels

“…For validation, we test over synthetic data sets such as the Illinois 200-bus system and South Carolina 500-bus system [31], [54]. We also test our results with realistic data from our utility partners.…”

“…Further, for each event type, we randomly create 2 different locations in the system. Similar treatments are implemented in many related studies [14], [15], [23], [31], [36].…”

“…We utilize Illinois 200-bus system and South Carolina 500bus system [31] to generate event data. Five event types are considered, including trip, generator trip, single-phase-toground fault, phase-to-phase fault, and three-phase fault.…”

“…For the numerical verification, the KTDC-Se method is tested extensively at various conditions on the Illinois 200bus system, South Carolina 500-bus system [31], and realistic data sets from our utility partners. These conditions include different loading conditions and PMU penetrations, etc.…”

Fast Algorithms for Estimating the Disturbance Inception Time in Power Systems Based on Time Series of Instantaneous Values of Current and Voltage with a High Sampling Rate

“…The training was carried out on PMU measurements before, after and during a disturbance: current magnitudes and phases, positive, negative and zero sequences. The problem of emergency control based on machine learning was considered in [52]. The problem itself boiled down to insufficient datasets for training in real power systems.…”