Review of the Application of Deep Reinforcement Learning in Multi-Component Coordinated Control of Microgrids

Microgrid plays an important role in distributed new energy consumption, load optimization, and improving energy efficiency. However, the intermittency of new energy output and the randomness of load-side power consumption behavior make the microgrid a dynamic complex system, and it is difficult to establish its accurate mathematical model, which brings great challenges to the optimal operation of the microgrid. Deep reinforcement learning (DRL) finds the optimal strategy through trial and error interaction with the environment, does not rely on accurate modeling of micro source output and load, is suitable for solving sequential decision-making problems, and has advantages in solving microgrid optimization operation problems with a large number of uncertainty units. Therefore, this paper reviews and analyzes the modeling and optimization of microgrids, the principle of DRL, the application of DRL in microgrids, and the application of DRL in the coordinated operation of multiple components, and finally prospects the algorithm interpretability, reward function setting, and user privacy faced in the applications.