A Modulated Model-Free Predictive Control of a Double-Stage AC-DC Converter for Energy Storage Applications

This paper presents a modulated model-free pre-dictive control (MMFPC) for a double-stage ac-dc converter integrating a vanadium redox flow battery (VRFB) with the electrical grid. The proposed control scheme enables simultaneous regulation of the active and reactive power exchanged with the three-phase low voltage ac (LVAC) grid, the de-link voltage between the converter stages, and the battery current, through a single cost function optimization. Compared to traditional model-based approaches, the converter is here considered as a black box, with system dynamics identified using an auto-regressive with exogenous input (ARX) model, based solely on input-output measurement data. The simulation results demonstrate that the proposed MMFPC achieves dynamic performance comparable to a state-of-the-art modulated model predictive control (MMPC), while eliminating the need for detailed system modeling and parameter tuning. Additionally, the MMFPC exhibits superior robustness, significantly reducing tracking errors under model uncertainties and parameter variations.