Renewable energy integration and microgrid energy trading using multi-agent deep reinforcement learning

To reduce global greenhouse gas emissions, the world must find intelligent solutions to maximise utilisation of carbon-free renewable energy sources. In this paper, multi-agent reinforcement learning is used control a microgrid in mixed cooperative and competitive setting. The agents observe fluctuating demand, dynamic wholesale prices, intermittent sources hybrid storage system renewables costs grid. addition, an aggregator agent trades with external microgrids competing against one another their own bills. For this, algorithm deep deterministic policy gradients (DDPG) DDPG (MADDPG) are compare use single controller versus multiple distributed agents, along variants distributional (D3PG) twin delayed (TD3). research found it significantly more profitable for primary sell on its terms rather than selling back utility grid, also beneficial as they methods that produced greatest profits were approaches where each has reward function based principle marginal contribution from game theory. better able evaluate performance controlling individual components environment which allowed them develop unique policies different types system.