Optimal load-sharing in isolated DC microgrids using an adaptive model predictive control

This paper presents an adaptive one-step model predictive control (MPC) strategy for optimal load-sharing and voltage regulation in isolated direct current (DC) microgrids. Utilizing a port-Hamiltonian representation within a reduced model, the proposed secondary control method ensures physically consistent actions and incorporates a recursive least-squares algorithm for real-time estimation of the reduced conductance matrix. The one-step formulation enables the analytical derivation of the optimal control law, ensuring convexity and global optimality for real-time implementation. Additionally, a convex optimization procedure is introduced to compute droop control gains, maintaining the passivity properties of the system while balancing current sharing and voltage regulation within operational constraints. Dynamic simulations on a realistic DC microgrid benchmark validate the method’s effectiveness, demonstrating enhancements in load-sharing accuracy, voltage regulation, and responsiveness to abrupt demand changes. These results highlight the potential of the proposed MPC framework to improve the reliability and efficiency of DC microgrid operations.