Experimental Realization of Koopman-Model Predictive Control for an AC-DC Converter

This paper experimentally demonstrates the Koopman-Model Predictive Control (K-MPC) for a real AC-DC converter. The converter is typically modeled with a nonlinear time-variant plant. We introduce a new dynamical approach to lifting measurable dynamics from the plant and constructing a linear time-invariant model that is consistent with control objectives of the converter. We show that the lifting approach, combined with the K-MPC controller, performs well across the full experimental system and outperforms existing control strategies in terms of both steady-state and transient responses.

💡 Research Summary

This paper presents the first experimental demonstration of Koopman‑Model Predictive Control (K‑MPC) applied to a real single‑phase AC‑DC full‑bridge boost rectifier. The converter, driven by SiC MOSFETs and supplied by a 50 Hz sinusoidal voltage (28√2 V amplitude), feeds a resistive load (G = 0.01 S) and a constant‑power load (P = 25 W). The control objectives are twofold: (1) maintain the DC‑side voltage mean at a desired value Vd = 48 V, and (2) enforce a unity power factor on the AC side, i.e., the AC current should be a pure sinusoid Id sin ωt with constant amplitude Id. From the power‑balance equation, Id is theoretically 2.44 A. Additionally, the AC current is limited to 1.5 Id ≈ 4 A to protect the MOSFETs.

The core contribution lies in constructing a linear time‑invariant (LTI) Koopman‑based model (KOM) from measured data using a dynamic lifting map rather than the conventional static observable dictionary. The authors adopt the Generalized State‑Space Averaging (GSSA) technique to compute moving harmonic averages of the measured current and voltage over each AC period. Specifically, the lifted state vector is defined as

z