Self-Tuning Finite-State Model Predictive Control with Quasi-Passive Grid-Impedance Estimation via Active and Reactive Power Modulation

This paper proposes a Finite Control Set Model Predictive Control (FCS-MPC) algorithm for a grid-connected three-phase inverter with an LCL filter, enhanced by an online grid impedance estimation method. The estimation relies on controlled active and reactive power variations, from which the complete RL impedance at the point of common coupling (PCC) is derived. Since the measurements are taken across the filter capacitor and the grid-side inductor of the LCL filter, the estimated impedance naturally includes both grid and grid-side filter contributions; these are compensated in real time using the known parameters R2 and L2 to obtain the actual grid resistance and inductance. The updated impedance values are directly integrated into the MPC prediction model to improve power tracking and harmonic rejection under varying grid conditions. A dedicated trigger logic based on grid voltage deviation and phase-A current THD activates the estimation only when significant impedance changes occur, avoiding unnecessary perturbations. The method has been validated through simulations, hardware-in-the-loop (HIL) experiments, and preliminary real-grid tests, showing accurate and robust impedance tracking, self-validation capability after incorrect estimations, and substantial improvement of current quality during abrupt grid parameter variations.