LOOP-MAC: Learning-Based Multi-Agent Coordination with Guaranteed Local Feasibility in a Virtual Power Plant
LOOP-MAC is a decentralized learning framework for coordinating many small energy sources (DERs) into a single Virtual Power Plant (VPP). It uses neural networks to speed up coordination, ensuring local feasibility by construction. This method significantly reduces the gap between the desired power commitment and what is actually achievable, improving efficiency compared to traditional methods.