Aiming at the problem that DC bus voltage is susceptible to PV output power changes and load disturbances
an Active Disturbance Rejection Control strategy based on conditional entropy and the deep reinforcement learning Soft Actor-Critic (SAC) algorithm is designed. By introducing the conditional entropy as an evaluation index to measure the degree of perturbation of the system and combining it with the deep reinforcement learning algorithm
real-time self-driven optimal tuning of the controller bandwidth is carried out to ensure that the energy storage unit containing a DC/DC converter can suppress the bus voltage fluctuation and improve the safety and stability performance of the DC microgrid. Finally
the feasibility of the proposed strategy is demonstrated by comparing it with the double-closed-loop PI control and the traditional linear active disturbance rejection control under different operating conditions.
HAARNOJA T, ZHOU A, ABBEEL P, et al.Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor.[C]//International conference on machine learning. Pmlr, 2018: 1861-1870.