Towards Distributed Learning of PMU Data: A Federated Learning Based Event Classification Approach

Abstract

Phasor Measurement Units (PMUs) stream mission-critical high-frequency voltage and current waveforms. We present a privacy-preserving federated deep learning framework allowing independent utility operators to train high-accuracy grid event classification networks without sharing proprietary telemetry, maintaining 88-90% accuracy across 3,877 real-world events.

@inproceedings{sajjadi2023pesgm,
  title     = {Towards Distributed Learning of {PMU} Data: A Federated Learning Based Event Classification Approach},
  author    = {Sajjadi Mohammadabadi, Seyed Mahmoud and Liu, Yunchuan and Canafe, Abraham and Yang, Lei},
  booktitle = {2023 IEEE Power & Energy Society General Meeting (PESGM)},
  pages     = {1--5},
  year      = {2023},
  organization = {IEEE},
  doi       = {10.1109/PESGM52003.2023.10252874}
}

Author Profile