Microgrid Assessment And Ml-Based Power System Faults Detection Leveraging Real-Time Co-Simulation

The rapid growth of distributed energy resources (DERs) and the increasing reliance on data-driven decision making have reshaped the operational challenges of modern electric power systems. As microgrids become more prominent in distribution networks, utilities require methods that unify planning, control, and real-time situational awareness to ensure resilient operation under faulted or uncertain conditions. The goal of this MSEE thesis is to design and validate a latency-aware ML framework for rapid, reliable fault detection in distribution grids. To achieve the goal of the thesis, there are three specific objectives. Objective 1 evaluates optimized microgrid configurations under varying DER levels and geographic locations, quantifying annual operational cost, emissions, renewable penetration, and levelized cost of energy. Objective 2 examines microgrid dynamic behavior during disturbance-driven islanding events by integrating DERs, automatic voltage regulation (AVR), automatic generation control (AGC), and other grid-supportive control mechanisms in a synthetic transmission environment. Objective 3 develops and benchmarks PMU-based ML models, including supervised learning (SL), unsupervised learning (UL), and reinforcement learning (RL) to detect faults across diverse scenarios and prepares some of these models for real-time deployment. The thesis consists of six chapters. Introduction and literature review are presented in Chapters 1 and 2. Chapter 3 contributes to developing a stability-focused evaluation of DER-rich microgrids, demonstrating how coordinated control strategies significantly enhance voltage and frequency performance during islanding events. Chapter 4 contributes to developing a comparative assessment of ML-based fault detection algorithms using both synthetic and utility datasets, highlighting their trade-offs in accuracy, robustness, and generalization to unseen disturbances. Chapter 5 contributes to developing a real-time OpenDSS-OPAL-RT co-simulation framework that streams measurements into ML and RL models, enabling evaluation of detection accuracy, false-alarm behavior, and latency limitations under operational timing constraints. Major findings of the thesis are outlined in Chapter 6. This thesis advances state-of-the-art by introducing a region-aware microgrid feasibility framework that incorporates local environmental conditions, resource availability, and operational constraints including the demonstration of hydrokinetics as a viable DER option in HOMER Pro software. It further integrates DER dispatch with AVR and AGC to evaluate microgrid stability during severe disturbance, showing that coordinated control substantially enhances frequency and voltage resilience during islanding. Furthermore, this thesis also delivers comparative analysis of SL, UL, and RL models for PMU-based fault detection in power systems using real-world data, revealing the adaptability of RL under evolving grid conditions. Finally, it presents real-time evaluations of SL and RL models within an OpenDSS–OPAL-RT co-simulation, demonstrating that ML-based detectors can operate effectively under real-time latency and sampling constraints and establishing a foundation for future real-time fault classification and mitigation.