Sparse Data-Driven Modeling and Optimal DC Voltage Control of MTDC System

This paper presents a novel data-driven model predictive control (MPC) strategy for optimal DC voltage regulation in multi-terminal high-voltage direct-current (MTDC) systems based on voltage source converters (VSCs). The proposed strategy utilizes the sparse identification of nonlinear dynamics with control (SINDYc) method to develop a data-driven model that accurately captures the nonlinear dynamics of grid-side VSCs. Pseudo-random binary signals (PRBSs) are used to collect training data for model development. The accuracy and robustness of the SINDYc model are validated under various training conditions and compared with conventional data-driven methods. Based on the SINDYc model, an MPC algorithm is formulated to minimize DC voltage deviations and restore the voltage to nominal operating points. Comparative case studies are performed to demonstrate that the proposed MPC strategy significantly enhances the effectiveness and stability of real-time DC voltage regulation in MTDC systems, particularly under time-varying wind power inputs.