Data-Driven Adaptive Unscented Kalman Filter for Time-Varying Inertia and Damping Estimation of Utility-Scale IBRs Considering Current Limiter

Grid-forming inverters, such as the virtual synchronous generator (VSG), can emulate constant or time-varying inertia to mitigate frequency stability issues. This paper proposes a data-driven variational Bayesian adaptive unscented Kalman filter (VBAUKF) to estimate the VSG-based inverter inertia and damping factor using its terminal measurements. By adopting the Thevenin equivalent idea, the virtual frequency of VSG is estimated first. Utilizing the estimated virtual frequency and considering the effects of the inverter current limiter, the time-varying inertia and damping factor estimation problem is reformulated into the state-space model-based dynamic state estimation framework. The measurements include the obtained virtual frequency, inverter terminal real, and reactive power while the unknowns are inverter inertia, damping factor, internal virtual rotor speed, and angle. To this end, an innovative VBAUKF is proposed with the advantages of dealing with unknown and time-varying models and measurement uncertainties. Numerical results on the modified IEEE 39-bus system and IEEE 118-bus power system demonstrate that the proposed estimator significantly outperforms other state-of-the-art approaches under various scenarios.