Date of Award
Fall 12-9-2025
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Electrical and Computer Engineering
First Advisor
Ayman El-Refaie
Second Advisor
Edwin Yaz
Third Advisor
Majeed Hayat
Abstract
Modern microgrids operate under conditions that differ significantly from traditional power systems, including high penetration of inverter-based resources, variable operating modes, bidirectional power flows, and communication-dependent monitoring. These characteristics undermine the reliability of conventional protection schemes based on fixed settings. This dissertation investigates the use of dynamic state estimation (DSE) as a foundation for adaptive, setting-less microgrid protection, enabling faster, more reliable decisions without relying on static thresholds or pre-calculated settings. A unified simulation and evaluation framework is developed to compare a broad family of DSE methods—including Kalman filter variants, robust H∞ filters, particle and ensemble filters, and moving-horizon estimation (MHE) — across a wide range of operating scenarios. The framework includes a Real-Time stack (L0–L5) that channels synchronized sensing through model-based estimation and settings-less validation to protection actuation with feedback, and an Experiment/Analysis stack (L6–L12) that orchestrates scenarios, curates data, conducts rigorous statistical evaluation and ablation, and yields deployment guidance alongside reproducible artifacts. Performance is analyzed using a consistent set of metrics capturing detection speed, fault selectivity, false-trip behavior, estimator health, robustness to noise and model mismatch, and communication dependency. The analysis incorporates realistic measurement conditions from both PMU and SCADA systems, along with variations in DER behavior, line parameters, load dynamics, communication impairments, and cybersecurity threats. Across more than 73,000 simulated events, the results show that DSE-based, settings-less protection can offer substantial improvements over traditional schemes. In particular, two-layer moving horizon estimation (TL-MHE) demonstrates consistently strong performance across diverse microgrid environments, balancing speed, reliability, and robustness to uncertainty. The dissertation concludes by providing high-level guidance on selecting suitable estimator families and protection architectures for different microgrid contexts, offering a practical roadmap for deploying adaptive protection in modern distribution networks.
Comments
Doctor of Philosophy (PhD)