Black-box Dynamic Model Identification for a Quadruple-Active-Bridge Converter

The isolated Quadruple Active Bridge (QAB) topology facilitates the flexible interconnection of multiple sources and loads in DC grids. However, this topology poses significant modelling challenges due to strong cross-coupling and multiple operating modes. This paper proposes a model-free, data-driven approach for open-loop characterization of the QAB as a Multiple-Input Multiple-Output (MIMO) system. Utilizing the Dynamic Mode Decomposition with Control (DMDc) as a blackbox identification technique, the system and input matrices are simultaneously identified from measurement data captured from the simulation. In order to address the nonlinear behaviours that occur across operating points, a polytopic model is constructed from a weighted combination of local linear models. The validity of the proposed method is demonstrated through circuit-level simulations and eigenvalue analysis under varying operating conditions, thereby evidencing both high accuracy and robustness to measurement noise.