A High-Fidelity Method for Identifying Grid-Forming Inverter Frequency Support Characteristics with Noise Robustness

Grid-Forming (GFM) inverters are critical for ensuring frequency stability in power systems with high penetration of renewable energy. Unlike conventional Grid-Following (GFL) control, GFM provides an intrinsic, coupled inertial and damping response, mimicking synchronous generators. This inherent coupling makes traditional decoupled assessment methods inaccurate. This paper proposes a high-fidelity, noise-robust system identification framework for validating GFM performance. The method is based on a “grey-box” model derived from Virtual Synchronous Machine (VSM) theory, which incorporates physical delays. A key innovation is the application of a Savitzky-Golay (SG) filter to preprocess noisy measurement data, enabling the accurate extraction of the Rate of Change of Frequency (RoCoF). The identification is formulated as a bound-constrained non-linear optimization problem to synchronously identify key physical parameters: the equivalent inertia constant (H), damping coefficient (Dp), and response delays. Validated on a real-time simulation platform, the proposed method achieves high accuracy even in high-noise environments. This represents an order-of-magnitude improvement in accuracy compared to “naive” identification methods that omit data preprocessing. The framework provides a reliable engineering tool for GFM grid-connection testing and model validation.