Composite Model Predictive Control Strategy for Triple-Active-Bridge-Based MMDTC-PET with Enhanced Disturbance Resilience

To address the delayed dynamic response and susceptibility to overcurrent surges of the triple-active-bridge (TAB)-based modular multilevel DC-link T-type converter power electronic transformer (MMDTC-PET) under load step transients, this paper proposes a composite model predictive control (MPC) strategy to enhance the system’s disturbance rejection capability. The proposed strategy adopts a hierarchical control architecture comprising a proportional-integral (PI) outer power loop and an MPC inner loop. To tackle the heavy computational burden caused by multi-port coupling, dimensionality reduction is performed by equating the phase-shift ratios of the two primary-side ports of the TAB, which effectively mitigates the computational cost. Furthermore, a dynamic overcurrent hard-constraint penalty term and a control-effort smoothing term are incorporated into the multi-objective cost function. This not only achieves steady-state error-free tracking of the output voltage but also enhances physical operational safety under complex conditions at the algorithmic level. Comparative simulation results under various operating conditions demonstrate that, compared with the traditional PI-based phase-shift control, the proposed strategy exhibits a faster dynamic response under load step conditions. Moreover, it suppresses transient power overshoots and overcurrent impacts during input voltage sags and recoveries, thereby significantly improving the transient disturbance rejection capability and operational robustness of the system.