Unified Co-Optimization of Converter Parameters and Switching Modulation for Enhanced ZVS in Low-Lm DC--AC DABs

Abstract

This article presents the harmonic superposition-based theoretical approach for analyzing the impact of low magnetizing inductance $L_{m}$ in extending the zero-voltage switching (ZVS) range in a DC–AC dual active bridge (DAB) converter. A comprehensive design optimization framework has been developed to obtain the optimal ratio of magnetizing inductance to power transfer inductance and the optimal transformer turns ratio. An exponential decay function accounting for the nonlinearity of $C_{oss}$ has been derived and utilized in formulating the ZVS inequality constraints, which are used to establish a multivariable, multiconstraint, and single-objective optimization framework to acquire the optimized control variables that enhance the average number of ZVS legs while co-optimizing the efficiency profile. Furthermore, the control strategy is implemented using a Fourier-based regression model, enabling high accuracy and faster real-time execution. Finally, a GaN-based 500-W/100-kHz low- $L_{m}$ DAB hardware prototype is developed and evaluated against the single inductor (SI) DAB model to validate the analytical framework and its effectiveness in extending the ZVS range. The experimental results for low- $L_{m}$ DAB demonstrate an efficiency improvement of 2.8% and 3.2% at 20% and 40% load, respectively, due to extended ZVS range and a slight efficiency decrease of 0.3% at full load when compared to single power-transfer inductor (SI-DAB).