Optimizing Operation and Control Strategies for Distributed Energy Resource-Integrated Microgrids: Evolutionary and Neuroevolutionary-Based Decentralized Primary Control in Islanded Mode

No Thumbnail Available

Date

2025-07

Journal Title

Journal ISSN

Volume Title

Publisher

Addis Ababa University

Abstract

The accelerating global climate crisis—combined with the uneven distribution of fossil fuel resources, rising oil prices, increasing demand for modern energy systems, and the global shift toward low-carbon economies—has necessitated a fundamental transformation in how energy is produced, distributed, and consumed. Renewable energy sources (RES) such as solar and wind are central to this transition. However, their integration into power systems presents persistent challenges due to intermittency, decentralization, and the limitations of traditional grid infrastructure. Microgrids (MGs) have emerged as a flexible and resilient solution for the decentralized integration of distributed energy resources (DER), playing a key role in advancing intelligent, low-carbon power networks. As the demand for sustainable, secure, and responsive energy systems grows globally, MGs—especially those incorporating DER and hydrogen technologies— are becoming indispensable components of the evolving energy landscape. While MGs hold promise for energy resilience and sustainability, their implementation faces challenges like high initial capital investment for infrastructure, technical complexity in integrating diverse DER, and interoperability issues with legacy grids. These barriers affect stable operation and the broader scalability of MG solutions. Accordingly, this research is directed toward addressing these multifaceted issues to enable more effective localized generation and control, advancing sustainability, energy independence, and resilience. The study begins by developing and evaluating integrated strategies to overcome the technical, economic, and environmental hurdles associated with MG deployment as a first objective. It focuses on the optimized design of a sustainable MG architecture that incorporates solar, wind, and hydrogen-based storage alongside sector coupling mechanisms. Through systemlevel analysis, the research underscores hydrogen’s pivotal role in increasing grid flexibility, compensating for the intermittency of renewable resources, and enhancing long-term energy security. This foundational investigation affirms the value of MGs in supporting decarbonization and improving energy resilience through hydrogen-enabled sector coupling and crosssectoral integration.In regions where utility grids are unreliable—characterized by frequent outages—MGs are often required to operate in islanded mode. While islanded operation enhances energy autonomy and resilience, it also introduces significant technical complexity. Without main grid support, islanded MGs must coordinate energy balance and economic operation across both short-term dynamics and long-term planning. Moreover, islanded MGs dominated by converter-interfaced DERs face considerable challenges in maintaining voltage and frequency stability under dynamic load and generation conditions. These operational challenges highlight the need for robust and well-coordinated control strategies specifically designed for islanded MG environments. In response, the subsequent phase of the thesis develops and evaluates advanced decentralized primary control strategies for islanded MGs. Within this framework, two distinct approaches are pursued: a virtual complex impedance-based method, which enhances transient response, system stability, and power-sharing accuracy by shaping the converter’s output impedance; and a computationally guided optimization framework, which systematically tunes control parameters to better coordinate with the dynamic characteristics of the interfacing filter and electrical network. These strategies are grounded in a detailed analysis of VSCs, including their control architectures and interfacing filters, which are vital for maintaining system stability and ensuring power quality. Building on these strategies, the final phase of the thesis introduces an Adaptive Hybrid PSO-Embedded Genetic Algorithm (AHPEGA) for the neuroevolutionary training of Multilayer Perceptron Controllers (MLPCs) in VSC-based islanded MG. This approach combines the global search efficiency ofGAwith the fine-tuning capabilities of PSO to dynamically optimize both the weights and biases, as well as hyperparameters, of the neural network—resulting in improved convergence and generalization in nonlinear control tasks. Overall, this thesis presents a comprehensive, multi-layered approach to MG control— encompassing sustainable system design and advanced decentralized control strategies— and delivers scalable, sustainable, and computationally guided solutions for resilient operation in islanded mode. These contributions establish a foundation for advancing intelligent control in islanded MGs integrated with renewable resources, thereby supporting the broader transition toward robust, efficient, and sustainable power systems.

Description

Keywords

Power grid, Microgrids, FCEVs, Electrolytic hydrogen, Emission reduction, Renewable energy curtailment, Islanded operation, Virtual complex impedance, Control parameter optimization, PSO-MLPNN, PSO-embedded GA, Neuroevolutionary training, Multilayer Perceptron Controllers (MLPCs)

Citation