Model-predictive control with admittance matrix estimation for the optimal power sharing in isolated DC microgrids

Direct current (DC) microgrids are relevant in modern energy systems due to their high efficiency, simplified architecture, and capability of direct integration of distributed resources (DERs). A proportional power-sharing is essential in these grids to balance the power injections according to the capability of each DER. However, the intermittency of DERs, load variations, and the non-linear nature of the model are major challenges. Therefore, this paper proposes a non-linear model predictive control (MPC) alongside an estimator for the nodal admittance matrix. By using MPC, it is possible to achieve optimal operation, considering voltage and power constraints. Moreover, the estimator enables the consideration of load variations with a reduced number of measurements. The robustness of the proposed control strategy is evaluated both in simulations and through a Power-Hardware-in-the-loop (PHIL) implementation. Radial and meshed microgrids were tested with different numbers of nodes. These results validate the practical feasibility and performance of the proposed approach.