MMC Equivalent Modeling Method of AC/DC Hybrid System Based on Model-Data Hybrid Drive

The conventional power system modeling methods have intrinsic disadvantages in representing unmodeled dynamics and ambiguous information. Specifically, the conventional modeling of converter valve usually ignores the high-frequency switching state. In order to break through the limitations of traditional modeling methods, this paper aims at giving an effective modeling strategy that can accurately describe the unmodeled dynamics and uncertainty of AC/DC hybrid interconnection systems. The validity of this method is enhanced by the model-data hybrid-driven modeling concept based on the digital twin. The proportional integral control link of voltage source converter (VSC) is modeled by the linearized mathematical method. The data-driven model of pulse width modulation (PWM) loop of VSC is trained and established by the time convolution neural network (TCN) algorithm, which can describe accurately the high-frequency switching state of switching device. The Simulation experiments on a real-world power grid demonstrate the high effectiveness of the proposed modeling method and the excellent performance of the hybrid-driven model.