Doctoral Dissertations
Date of Award
8-2025
Degree Type
Dissertation
Degree Name
Doctor of Philosophy
Major
Electrical Engineering
Major Professor
Hector Pulgar
Committee Members
Fangxing Li, Amir Sadovnik, Jin Zhao
Abstract
This dissertation addresses the challenges introduced by the rapid transformation of modern power grids, driven by the growing penetration of inverter-based resources, declining system inertia, and emerging technologies. These shifts introduce increased variability and uncertainty into power system dynamics, complicating the task of maintaining stability. This work explores how combining model-based strategies with data-informed techniques can enhance system control, assessment, and awareness, providing scalable tools to improve grid reliability.
The first contribution focuses on improving the coordination of damping controllers for low-frequency electromechanical oscillations. Traditional controllers are typically tuned for fixed operating conditions and dominant modes, limiting their robustness under varying scenarios. To overcome this limitation, a data-informed coordination strategy is proposed. By leveraging wide-area measurements, it enables real-time optimization of damping control actions without relying on linearized models. This decentralized, non-intrusive framework enhances the damping of complex, multi-mode oscillations across the grid.
The second and central contribution presents a high-fidelity and computationally efficient method for estimating system frequency response (SFR) and frequency nadir—key indicators of system resilience following large disturbances. Traditional methods typically trade accuracy for speed or vice versa, limiting their practical use for fast assessment or precise planning. To overcome these limitations, a novel modal-based estimation approach is developed, which exploits the system’s modal structure to predict the slow dynamic response that governs nadir behavior. This method accurately reconstructs the frequency trajectory from system modes and disturbance characteristics, enabling precise prediction of both nadir magnitude and timing. The approach is validated on standard IEEE test systems and the Salvadoran power grid, demonstrating significant accuracy improvements over legacy tools.
The final contribution introduces a hybrid model-AI framework for online estimation of SFR and frequency nadir. This approach integrates modal-based analysis with deep learning, enabling fast, predictive evaluation of major contingencies based on pre-contingency conditions and system-level features. The framework supports the identification of critical events and real-time forecasting of worst-case frequency deviations, enhancing operational awareness and decision-making.
Together, these contributions offer practical, hybrid model–data solutions for modern power system stability, addressing both current and future operational challenges.
Recommended Citation
Zelaya Arrazabal, Francisco, "Hybrid Model-Data Strategies to Improve Power System Control, Assessment, and Awareness. " PhD diss., University of Tennessee, 2025.
https://trace.tennessee.edu/utk_graddiss/12792