A Data‐Mechanism Fusion‐Driven Admittance Identification Method for Grid‐Connected Inverters Under Wide Grid Impedance

The output admittance model of grid‐connected inverter (GCIs) based on frequency sweeping is a critical tool for analysing the interaction characteristics between GCI and the grid. However, variations in grid impedance alter the GCI output admittance characteristics, making it impossible to apply a single admittance model derived under ideal grid conditions to scenarios with a wide range of grid impedances. Moreover, frequency sweeping operations under different grid impedances can lead to system instability near the critical short‐circuit ratio, rendering it difficult to determine the most crucial stability boundary and achieve stability prediction. To address this, this paper focuses on how to obtain the admittance model in unstable regions and proposes a data‐mechanism fusion‐driven approach based on artificial neural networks. By integrating the mechanism model with admittance sweep data under ideal grid conditions, an equivalent sequence admittance identification model is constructed, applicable to a wide range of grid impedances, while avoiding the risk of sweep‐induced resonance under critical stability conditions. Compared to data‐driven methods, the proposed method reduces root mean square errors of sequence admittance magnitude and phase identification by 77.73% and 48.65%, respectively. Simulations and hardware‐in‐the‐loop experiments validate the accuracy and applicability under wide grid impedance.