Abstract

Moving-discretized-control-set model predictive control is a powerful control scheme for dual-active-bridge (DAB) converter topologies, given its outstanding features in terms of fast dynamics and multiobjective optimization. However, the inaccuracy of the system model will degrade its performance, and the high-precision model of the DAB will also aggravate the complexity of controller design. To circumvent this issue, a model-free moving-discretized-control-set predictive control (MF-MDCSPC) scheme is first proposed for three-phase DAB (3p-DAB) converters in this article. First, a novel data-driven model is proposed to replace the traditional mathematical model of the 3p-DAB, which is built with the control input and output variables of the 3p-DAB. Next, an adaptive forgetting factor recursive least square algorithm is developed to estimate the parameters of the data-driven model, and a novel MF-MDCSPC based on the data-driven model is designed to enhance the parameter robustness of the 3p-DAB. Furthermore, a multilayer recursive method is presented to estimate the parameters of the proposed data-driven model in the predictive horizon, and an improved adaptive step mechanism is developed to realize a fast transition. Finally, comprehensive simulation and experimental results from a down-scale laboratory prototype are provided to verify the theoretic analysis and the efficiency of the proposed methodology.