الفهرس 
AcknowledgementOnly 14 pages are availabe for public view
 |  | from | 191 |  |  |
| | | from | 191 | | |
Abstract
”With the rapid development of electrical systems in recent years, microgrids (MGs) have become increasingly prevalent. The MG is a small-scale system that provides local energy production and storage capabilities, allowing for the seamless integration of various distributed energy resources (DERs), such as inverter-based DERs (I-DERs) and synchronous-based DERs (S-DERs). This integration yields numerous benefits, including increased power system efficiency, reliability, and sustainability. However, the decentralized nature of MGs and the incorporation of various types of DERs pose unique challenges to existing network protection schemes, necessitating the development of novel approaches to ensure reliable and secure operation.To investigate these issues, this thesis first presents a comparative framework that evaluates the complexities that MGs introduce to traditional protective relays. The thesis then reviews previous research on AC-MG protection schemes, including theoretical and practical implementations. This review critically assesses the effectiveness of various protection strategies, ranging from traditional methods to advanced methodologies that employ signal processing and machine learning techniques. Given the reliability of directional overcurrent relays (DOCRs) in considering fault current magnitude and direction before initiating trip commands, this thesis focuses on optimizing DOCR settings in MGs, addressing a wide range of issues and challenges.Although DOCRs are extremely beneficial in MGs, coordinating them in DER-rich grids is difficult due to the increased fault currents from DERs, particularly synchronous ones. This reduces the operating time of DOCRs with inverse characteristics and prevents primary-backup relay coordination. Several optimization algorithms have been used in the literature to define DOCR settings in MGs while accounting for a variety of constraints. Despite their contributions to DOCR coordination, these studies ignored DER transient stability and assumed post-fault stability. This assumption is problematic in MGs because DERs are susceptible to instability after fault clearance. Unlike large synchronous machines found in transmission systems, DERs in MGs have low inertia and damping, making transient stability considerations critical for preventing instability at post-faults. As a result, this thesis proposes a novel approach that employs shifted user-defined characteristics (non-standard characteristics) of DOCRs to ensure relay coordination and DER transient stability. Critical clearing times of DERs are combined with traditional coordination constraints to define optimal DOCR settings using genetic algorithm. The proposed scheme’s effectiveness is thoroughly evaluated on a modified IEEE 33-bus test system with various DER types, using DigSILENT and MATLAB software. Initially, DOCR settings are defined considering the transient stability of S-DERs only while also examining the impact of variable power production during off-peak periods and transformers’ inrush currents on the proposed settings. The coordination among DOCRs, auto-reclosers, and fuses is also discussed. Furthermore, this study investigates the effect of integrating I-DERs on transient stability and DOCR settings, given their susceptibility to instability after faults due to a lack of inherent damping and inertia when compared to S-DERs. Finally, an adaptive DOCR scheme based on unsupervised machine learning is proposed to handle a wide range of MG operational scenarios. This method divides network potential operating cases into groups that correspond to available commercial DOCR settings, with optimal settings defined and stored within the DOCRs for seamless activation when needed. Given that DOCRs rely on current and voltage measurements as operating and polarizing quantities, respectively, to detect and trip forward faults, they are susceptible to manipulation by attackers. Such manipulation may cause DOCRs to misinterpret fault conditions and generate false trip signals. Thus, to improve the resilience and security of DOCRs against various cyber-attacks, the latter part of this thesis introduces a novel scheme for detecting various cyber-attacks. The proposed scheme employs supervised machine learning to distinguish between genuine faults and malicious activity, allowing for more informed decisions on whether to permit or prohibit DOCR operations.The findings of this thesis emphasize the significance of the proposed strategies in protecting AC-MGs, ensuring system transient stability following faults, and improving the resilience and security of the proposed protection system against cyber threats.