Abstract：The integration of new energy units on a large scale has introduced numerous adverse effects on the

stability of system operation. Analysis of mechanisms and identification of impedance characteristics in gridconnected inverter（GCI）for new energy units are key to effectively addressing these issues. Therefore，a model and data-driven nonlinear modeling and impedance identification approach for GCI was presented. Firstly，

considering the influence of phase-locked loop dynamics，a small signal modeling was performed on the GCI of

new energy units，and a nonlinear functional relationship between the input and output variables of GCI impedance

identification was established. Secondly，a simulation model of the new energy units grid connected system by

Matlab/Simulink was built，and the dataset required for eXtreme gradient boosting（XGBoost）training under

various operating conditions was obtained. Then，simulation data was adopted to train XGBoost，and particle

swarm optimization（PSO）was employed to optimize the hyperparameters of XGBoost. Finally，the impedance

characteristics of GCI were scanned using RT-LAB hardware-in-the-loop testing technique combined with small

disturbance method to obtain the actual impedance values required for model validation. The effectiveness and

superiority comparison experiments show that the PSO-XGBoost model has higher GCI impedance identification

accuracy compared to other models.

      Key words：new energy units；grid-connected inverter（GCI）；impedance identification；eXtreme gradient

boosting（XGBoost）；particle swarm optimization（PSO）