Abstract:This study presents an AI-driven swarm-based control architecture for real-time optimization ofmicrogrid energy management. The proposed system integrates Particle Swarm Optimization(PSO), Ant Colony Optimization (ACO), and Reinforcement Learning (RL) to achieveintelligent and adaptive decision-making in a distributed energy environment. The microgridconsists of a 50 kW solar photovoltaic system, a 200 kWh battery storage unit, and a peakdemand of 1 MW. The hybrid model dynamically manages energy generation, storage, andconsumption to improve overall system performance. Simulation results demonstrate that theproposed approach significantly outperforms conventional methods. It reduces energy losses byapproximately 30%, achieving around 5% loss compared to 15% in traditional systems.Additionally, the model achieves a 25% reduction in operational costs, enhances grid stability by40%, and improves decision-making speed by reducing response time from 250 ms to 150 ms.Furthermore, the system minimizes grid dependency during peak demand periods and ensuresefficient utilization of battery storage, offering a scalable and cost-effective solution.
Sudhit Padakanti (Wed,) studied this question.