Data-driven modelling of grid-connected virtual synchronous machine dynamics using optimized dynamic mode decomposition and SINDy

In recent years, power grids have undergone rapid integration of converter-based power sources. Modern switching converters are replacing conventional synchronous machines. Owing to their inherent inertial property, virtual inertial (VI)-based virtual synchronous machines (VSMs) are increasingly being used in grid-connected converter control strategies With the availability of high-resolution measurement data from hardware measurements or simulations, the use of data-driven methods (DDM) for modelling system dynamics has become more popular than conventional modelling methods. This study employs a data-driven approach to model the dynamics of grid-connected VSM (GCVSM)-based converters. A simulation of the system’s detail switching model is used to create the DDM model, done by using Optimized Dynamic Mode Decomposition (ODMD) and sparse identification of nonlinear dynamical systems (SINDy). The results are compared with those of the actual switching model, the average nonlinear model, and the linearized model of the GCVSM. The ODMD performed very well, with a goodness of fit of approximately 89%, which is very close to that provided by the nonlinear model, which is approximately 94% for the 50[Formula: see text] active power setpoint scenario. The data-driven model developed in this study is a reduced-order model that can be useful for understanding the transient and steady-state characteristics of systems and for developing control strategies.