Impact of PLL Design on Data-Driven Models for Grid-Connected Single-Phase Inverters

This paper investigates the impact of quadrature signal generation based phase-locked loop (QSG-PLL) methods on data-driven modeling of grid-connected single-phase inverters (GCSI). The magnitudes of the grid voltage and current injected by a GCSI simulated in MATLAB/Simulink are estimated by each QSG-PLL approach. The best transfer function for the GCSI model is estimated in terms of goodness of fit, using the magnitudes obtained by each approach and the instrument variable system identification approach. Different grid disturbances (e.g., voltage sag, phase angle jumps, harmonic distortions, frequency fluctuation) are used to evaluate the performance of each data-driven model in comparison to the simulated model. The paper provides researchers guidance on which QSG-PLL to use to model GCSIs using data-driven methods.