Event-Triggered Communication Mechanism for DRL Secondary Frequency Compensation in AC Islanded Microgrid

Frequency deviation challenges the stable operation of AC islanded microgrids, and the development of machine learning-based control schemes offer a new solution to this problem. This paper proposes a deep reinforcement learning (DRL) secondary frequency control scheme for islanded microgrids that contains an event-triggered communication link. Compared to the traditional secondary control scheme, the DRL-based scheme can provide more accurate frequency regulation with strong adaptive capability. Simultaneously, the DRL-based solution will bring a larger computational burden for the controller. An event-triggered communication mechanism based on sampled data is also proposed. This mechanism activates the secondary control process only upon the occurrence of an event trigger, significantly reducing the computational load for the intelligent secondary control system. The validity of the proposed solution is verified using Matlab/Simulink, and its effectiveness is further confirmed through real-time simulation on the OPAL-RT platform.