Enhancing resilience of inverter-based resources: Adaptive neural network control for virtual synchronous generator damping under grid impedance uncertainty

The increasing integration of renewable energy sources (RESs) into the power grid has led to the widespread replacement of synchronous generators (SGs) with inverter-based resources (IBRs). Unlike SGs, inverters lack inherent inertia, which is critical for frequency stability. Virtual synchronous generator (VSG) control has emerged as a promising solution by emulating SG dynamics in inverter control. However, conventional VSGs rely on fixed damping and inertia parameters designed for specific operating conditions $-$ typically assuming a fixed grid strength. These assumptions often fail when operating conditions change and the impedance of the grid varies, leading to degraded performance or even instability. To address these limitations, this paper proposes a data-driven neural network (NN)-based damping controller that replaces fixed damping terms with adaptive measurement-driven damping. The NN is trained offline across a wide range of short-circuit ratios (SCRs) and $X_g/R_g$ impedance profiles, allowing it to generalize across operating points without requiring online re-tuning. A Lyapunov-based input-to-state stability (ISS) analysis is presented to guarantee system stability under bounded disturbances, even in the absence of an explicit transfer function. Finally, real-time simulations validate the proposed controller’s damping performance under both strong and weak grid conditions, and across different impedance types.