Real-time Energy Management of Microgrid Using Reinforcement Learning

Abstract

Driven by the development and application of smart grid and renewable energy sources (RES) generation technologies, microgrid (MG) plays an important role in environmental protection and optimization of the grid structure by integrating local loads and distributed energy. However, the stochastic and intermittent nature of RES have caused difficulties in the economic energy dispatching of MG. Inspired by reinforcement learning (RL) algorithms, this paper proposes a novel learning-based control MG scheduling strategy. Unlike traditional model-based methods that require predictors to estimate stochastic variables with uncertainties, the proposed solution does not require an explicit model. The proposed method is simulated in the environment composed of realistic data, and the effectiveness of the method is explained and verified.