Neural Differential Equations-Driven Predictive Control for Improving Microgrids Stability With Satellite Internet

Remote microgrids with satellite internet provide a viable solution for integrating renewable energies. However, their stable operation is vulnerable to the communication services quality. Noises from satellite internet deteriorate the stable control performance and slow the restorage of islanded microgrid after large disturbances. This work proposes a neural differential equation-driven predictive control (SNPC) strategy to coordinate the grid-forming converters to enhance the microgrid's transient stability under signal noises. The neural differential equations are established as the prediction model that robust to the control mode-switch and signal noises. Integrating the interpretability of model-driven approaches with the flexibility of data-driven methods, the proposed SNPC enables intuitive prediction of the microgrid's center of inertia trajectory and the dynamic behaviors of converters. Each stable control signal is optimized using the converters’ dynamic trajectories and system trends. This thus accelerates the convergence of islanded microgrids to an asymptotically stable point and enhances the transient stability. Simulations and hardware-in-the-loop experimental validations are utilized using actual satellite internet latency for an IEEE 38-bus system and a practical islanded microgrid-based system to demonstrate the effectiveness of the proposed SNPC.