A Model-Free Predictive Controller for Networked Microgrids with Random Communication Delays

This paper introduces a model-free predictive controller (M-fPC) to allow delay compensation carried out locally in networked microgrids. As the prediction horizon of the existing model predictive controllers (MPC) is limited by communication delays, the proposed controller offers a robust performance and accurate prediction even under large random delayed measurements. Its design philosophy is leveraged by constructing a prediction policy using the inner control loop dynamics in DC networked microgrids. By doing so, the said policy is able to administer the equivalent delay and construct a delay compensation signal accordingly. Finally, its robustness under various communication delay has been simulated under many disturbances and validated experimentally.