A Data-Driven Energy Management Strategy Based on Deep Reinforcement Learning for Microgrid Systems

Due to the interactions among schedulable equipment and the uncertainty of microgrid (MG) systems, it becomes increasingly difficult to establish accurate mathematical models for energy management. To improve the stability and economy of MGs, a data-driven energy management strategy must be proposed. In this paper, distributed generators (DGs) and an energy storage system (ESS) are taken as the control objects, and a data-driven energy management strategy based on prioritized experience replay soft actor-critic (PERSAC) is proposed for MGs. First, we construct an MG energy management model with the objective of minimizing the operation cost. Second, the energy management model is formulated as a Markov decision process (MDP), and the PERSAC algorithm is used to solve the MDP. Moreover, the sampling rule of the training process is optimized by using the prioritized empirical replay (PER) method. The analysis of numerical examples proves the effectiveness and practicability of the algorithm. By controlling DGs and the ESS, the operation cost of the proposed algorithm is the lowest compared with other algorithms.

Automated summary

In this paper, a data-driven energy management strategy based on prioritized experience replay soft actor-critic (PERSAC) is proposed for microgrid (MG) systems. The algorithm optimizes the operation cost by controlling distributed generators (DGs) and an energy storage system (ESS). Numerical examples show that the proposed algorithm achieves the lowest operation cost compared to other algorithms.