Distributed Secondary Model-Free Predictive Control Scheme for Droop-Controlled Isolated Microgrids

Distributed model-based predictive controllers are widely used as secondary-level controllers for microgrids. That is because they include the microgrid constraints in the problem formulation, have low computational time, and are versatile. However, these controllers have a disadvantage: They need to know the physical model of the microgrid to predict the evolution of the microgrid effectively. To solve this issue, different estimation methods have been developed to partially or entirely eliminate the controller model s dependence on the system knowledge. This approach, denominated model-free predictive controllers, has not been implemented on microgrids to the authors knowledge; therefore, this work proposes a distributed model-free predictive secondary level control scheme for microgrids that partially replaces the models of reported distributed model-based predictive control with autoregressive models. The proposed controller and its autoregressive model are tested through simulation; its performance is compared with a reported model-based predictive controller. The results show that the proposed model-free predictive controller can fulfil all secondary-level control objectives, even if the value of the passive components of the microgrid changes, in contrast to the modelbased predictive controller, whose performance decreases with the change in the passive component value. © 2025 IEEE.