Control-oriented unified modeling and hierarchical control of a cascaded DAB–Čuk converter for bidirectional battery charging

This paper presents a control-oriented modeling and validation study of a cascaded Dual Active Bridge (DAB)–Čuk converter for bidirectional battery-charging applications. The proposed topology combines the galvanic isolation and bidirectional power-transfer capability of the DAB stage with the continuous-current behavior and output-conditioning capability of the Čuk stage. Because the two stages are coupled through a finite intermediate DC-link capacitor, variations in DAB phase shift and Čuk duty ratio jointly influence the converter dynamics and complicate closed-loop design. To address this issue, a unified large-signal averaged model is first developed and then linearized to obtain a six-state small-signal representation of the cascaded system. The resulting model captures the dynamic interaction between the isolated front-end and the downstream DC–DC stage and provides a control-oriented basis for hierarchical regulation of the DC-link and battery-side variables. On this basis, a dual-loop control structure is implemented, in which the inner loop regulates the intermediate DC-link voltage through single-phase-shift (SPS) modulation and the outer loop regulates the battery-side current through duty-ratio control of the Čuk stage. The modeling and control framework is evaluated using MATLAB/Simulink studies and laboratory-scale experimental tests under reported bidirectional operating conditions. At the nominal G2V operating point, the analytical/simulation/experimental values of the intermediate DC-link voltage were 30.0/30.0/29.5 V, the battery-side voltage was 24.0/24.1/23.8 V, and the leakage-current peak was 9.0/9.2/8.5 A. These results provide local experimental support for the proposed unified modeling framework as a control-oriented description of the cascaded converter under the tested laboratory conditions, while broader transient validation, wider operating-range assessment, and parasitic-aware modeling remain subjects for future work.