@inproceedings{80bc1f32d879436dbd180c529f1defe8,
title = "A Machine Learning Framework for Identifying and Modeling Speed Governing Deficiencies in Hydropower Units",
abstract = "A common notion is that all hydropower units have active speed governing capabilities. However, this capability can be deactivated due to varying reasons. To uncover and quantify the existence of such units in the Western Electricity Coordination Council (WECC), this paper performs a data-driven analysis utilizing open-source and restricted datasets. A decision tree ensemble-based machine learning algorithm is designed to estimate governor models for the identified units. After updating the WECC power system files with estimated governors the impact on frequency profile is quantified. Results indicate that there are 182 hydropower units in the WECC PSS/E format power system dynamic data file that do not have governor models (representing their in-field speed regulating incapability). 74\% of these units are less than or equal to 20MW. It was observed that an additional 150MW to 200MW of response could be added which improved frequency nadir and frequency sensitivity.",
keywords = "classification, decision tree, frequency control, governor, Hydropower",
author = "Soumyadeep Nag and Richard Tapia and Alam, \{S. M.Shafiul\}",
note = "Publisher Copyright: {\textcopyright} 2026 IEEE.; 2026 IEEE Green Technologies Conference, GreenTech 2026 ; Conference date: 25-03-2026 Through 27-03-2026",
year = "2026",
doi = "10.1109/GreenTech68285.2026.11471622",
language = "English",
series = "IEEE Green Technologies Conference",
publisher = "IEEE Computer Society",
pages = "482--487",
booktitle = "2026 IEEE Green Technologies Conference, GreenTech 2026",
}