Microgrid Power Sharing: Adaptive vs. Nonlinear Predictive Models

Model predictive control (MPC) is an advanced optimization framework used in a closed-loop system to manage distributed energy resources (DERs). This study explores adaptive model predictive control with successive linearization (ADMPC-SL) and nonlinear MPC (NLMPC) for microgrid control, focusing on power sharing and load management. The microgrid model uses a centralized MPC with linear time-varying and nonlinear prediction models for precise demand fulfillment. This method enhances power distribution efficiency and voltage regulation while optimizing the setpoints of inverter-based devices. Real-time simulations conducted on a 2-DER islanded microgrid with OPAL-RT technologies confirm the effectiveness of the proposed method. Adaptive MPC with successive linearization addresses prediction errors from linearizing complex inverter models. Our results indicate that ADMC-SL requires 50% less computational effort than NLMPC, while NLMPC delivers superior accuracy.