Dynamically adaptable model Predictive Control with Signal Decomposed Synchronizer for Inverter Dominated Grids

Abstract

High penetration of power electronics in power generation is envisaged to revamp the legacy grid to Inverter Dominated Microgrids (IDM). Such a migration bringsforth unique challenges in grid synchronization and inverter power control arenas. An investigation on Variational Mode Decomposition (VMD) technique extended to grid synchronization of power converters in emerging non-linear grids is done. Number of modes and data fidelity parameter of VMD are tailored to achieve finer separation of the fundamental frequency in spite of the spectral band variations and minor frequency deviations of the voltage. The test cases are formulated based on CIGRÈ and IEEE Task Force 1159.2 repository on the foreseen power quality issues in the emerging power utilities. Two statistical indices, absolute error in percentage and RMS error, are chosen to assess the compliance of extracted signals with the applied values at cycle as well as sample level. A significant reduction in the tracking time in all the test conditions has been observed in simulation when compared to the conventional grid synchronizers. Appreciable accuracy levels are obtained in tracking the required signal attributes of the input signal. The frequency decomposition approach of VMD synchronizer exhibited an immunity towards random transient events like voltage surges, zero crossing disorders, phase and frequency jump, etc. Further, validation in hardware has been conducted with emulated grid signals representing a wide range of transient events and the results are found satisfactory. newlineNext, Model Predictive Control (MPC) has been identified as an upcoming converter control newlinethat can accomplish additional capabilities such as multi-mode operation, self-adaptability, high efficiency, seamless bi-directional power transfer, etc. besides the prime objective of reference tracking. But, implementing MPC in grid tied converter control demands its design parameters like sampling frequency, weight function coefficients and model parameters to be aptly chosen to match...