Robust Koopman EMPC for optimal frequency regulation of VSC-MTDC systems

This paper proposes a data-driven output feedback economic model predictive control (EMPC) strategy for load dispatch and frequency regulation in VSG multi-terminal HVDC (VSC-MTDC) systems. The deep Koopman operator is represented for modeling the nonlinear system. EMPC strategy is implemented to realize the economic load dispatch (ELD) and frequency regulation. To deal with the system uncertainties and obtain better control and optimization, a robust output feedback EMPC is utilized to solve the optimal laws for the feedback control and state observer. The asymptotic stability of the nominal closed-loop system is strictly ensured through the EMPC framework, and all trajectories converge uniformly to a predetermined neighborhood of the origin under uncertainty. The proposed method is verified through simulation of VSC-MTDC systems. The simulation results validate that the system achieves a more optimized load dispatch using this algorithm while enhancing the system’s robustness and stability. • A deep-learning Koopman architecture enables global linear embeddings for MTDC networks via spectral decomposition. • An EMPC strategy unifies ELD and frequency regulation, using output feedback to mitigate uncertainties effectively. • Rigorous proofs ensure closed-loop EMPC stability, guaranteeing convergence to an optimal equilibrium under disturbances.