Joint Online Adaptive Parameter Identification and Robust State Observation for Synchronous Generators

To address the state perception challenges of unmeasurable rotor inertia ( ) and vulnerability to topology mutations in low-inertia power systems, this paper proposes an Augmented-State Unscented Kalman Filter (AS-UKF) embedding grid algebraic equations. By analytically inverting the local impedance matrix to eliminate algebraic loops in differential-algebraic equations (DAE), a fifth-order pure differential adaptive model incorporating unknown is constructed. Time-domain simulations demonstrate: 1) a 98.69% noise suppression rate under normal fluctuations; 2) accurate cross-domain identification of using only electrical measurements (steady-state error <1%); 3) strong robustness against cascading faults (e.g., LVRT and topology mutations), restricting post-transient power angle errors to 2.77°. This method significantly enhances robust parameter perception and autonomous operation of generating units in complex grids.