Intelligent Energy Management in Microgrids Using Multi-Agent Reinforcement Learning

Abstract

The increasing integration of renewable energy sources and distributed energy resources has made smart microgrid energy management more complex due to renewable intermittency and dynamic operating conditions. This paper proposes a Multi-Agent Reinforcement Learning (MARL)-based energy management framework in which intelligent agents coordinate renewable generation, energy storage, controllable loads, and grid interaction. Through distributed learning and real-time decision-making, the framework optimizes energy scheduling, improves system reliability, reduces operational cost and carbon emissions, enhances renewable energy utilization, and provides a scalable, autonomous, and resilient solution for next-generation smart microgrids.