Energy Optimization Management of Multi-microgrid using Deep Reinforcement Learning

The paper investigates the energy optimization management problem for multi-microgrid (MMG) in the island case. MMG system could achieve greater consumption of distributed energy and a more stable power supply. While, due to the existence of uncertainties such as wind turbines, photovoltaics and loads, it is challenging to design an accurate energy optimization model to control power flow. Notably, conventional methods also appear to be inapplicable. To address the problem of MMG, a deep reinforcement learning optimal algorithm is applied in this paper. The simulation verifies that the proposed method could achieve the power balance within MMG and could minimize the total cost under uncertainties.