Online Optimization of Multi-Degree-of-Freedom Inductor Current RMS For Modular Multilevel Dual Active Bridge Converter

In order to improve the efficiency of the modular multilevel-neutral point clamped (MMC-NPC) three-level dual active bridge (DAB) dc-dc converter and to address the problem that optimization of multi-degree-of-freedom converters generally employs methods such as offline lookup tables. An online optimization strategy of multi-degree-of-freedom RMS inductor current of MMC-NPC three-level DAB dc-dc converter is proposed. Firstly, the optimization solution model of the dc-dc converter is established, and then it is solved by using the global optimal solver Gurobi, and the obtained results are put into the LSTM training, and the trained LSTM intelligences are equivalent to the fast agent prediction model, which can optimize the RMS value of the inductance current in the whole voltage and power ranges. Finally, the proposed online optimization strategy is validated in Matlab/Simulink simulation platform.