This paper proposes a systematic identification framework founded on the eigensystem realization (ER) to precisely model electronic power converters. The proposed framework furnishes an energy-based optimal reduction method to precisely identify the power converters' dynamics from simulated or actual raw data measured at the converter's ports. This approach does not require any prior knowledge of the topology or converter internal parameters to derive the system modal information. The proposed method's accuracy and feasibility are exhaustively evaluated via simulations and practical tests on a software-simulated and hardware-implemented dual activebridge (DAB) converter under steady-state and transient conditions. After different comparisons against the Fourier series-based generalized average model, the switching model, and experimental measurements; the proposed method attains a root mean square error less than 1% concerning the actual raw data and a computational effort reduction of 8.6 times regarding the Fourier-based model.