A rule-based selection algorithm for enhancing power quality in electrical distribution systems with microgrid controllers

The increasing integration of microgrids into electrical distribution systems presents opportunities and challenges in maintaining power quality (PQ) and grid stability, which are essential to ensure reliable operation, prevent equipment damage, and avoid service interruptions. Conventional control strategies often struggle to adapt to fluctuating PQ disturbances due to their limited flexibility and reliance on static parameters, which make them less responsive to dynamic and unpredictable grid conditions. This constrains their ability to mitigate issues such as current imbalances, where the current is unevenly distributed across the phases of the electrical system, and harmonic distortion, where current waveforms deviate from their ideal sinusoidal shape. This dissertation proposes a selection framework inspired by rule-based hyper-heuristics, justified by their flexibility and ability to generalize decision strategies across a wide range of PQ scenarios, making them suitable for dynamic and unpredictable environments. The framework regulates microgrid control actions by assigning optimized weights to a set of decision rules, enabling adaptive responses to fluctuating PQ disturbances. The selection scheme profits metaheuristic optimization techniques, specifically, the grey wolf optimizer, micro-Genetic algorithm, and the inertial version of particle swarm optimization to tune a rule-based decision-making mechanism. These metaheuristics were selected because of their reported effectiveness in optimizing power systems and industrial applications. The selection framework was evaluated using a single microgrid connected to a simplified low-voltage distribution model with residential loads operating at 120/240 V and 60 Hz, representing North American systems and subjected to close-to-reality PQ disturbances. The proposed framework was also rigorously evaluated across 90 PQ scenarios, demonstrating superior adaptability compared to standalone controllers. This adaptability is reflected in the model’s ability to maintain 93.33% of instances within the PQ thresholds ruled by international standards. Additionally, the rule-driven approach reduced around 80% of power losses provoked by current imbalances and harmonic distortions on average, translating to annual energy savings of up to 25,000 kWh (4,000 USD), reinforcing its economic and operational advantages. The findings of this dissertation suggest three potential research paths: scaling up to larger distribution systems by addressing computational complexity concerns, generalizing to diverse grid scenarios, and developing hybrid artificial intelligence approaches to refine decision-making processes. Therefore, this research contributes to the intersection of electrical engineering, control systems, and computer science by demonstrating how existing metaheuristic techniques can be integrated into a rule-based selection framework for decision-making under real-world scenarios. The computational contribution lies in the design, tuning, and validation of a cooperative control scheme that generalizes across diverse operating conditions, serving as a practical proof of concept for applying heuristic-driven decision models to dynamic energy systems.