Comparative Analysis of State and Parameter Estimation Techniques for Power System Frequency Dynamics

Abstract

Dynamic state and parameter estimation in current and future power systems are critical for advanced monitoring, control, and protection. There are numerous methods to perform dynamic state and parameter estimation; this paper compares the accuracy and computational time of four methods (i.e., Kalman filter (KF), extended Kalman filter (EKF), unscented Kalman filter (UKF), and moving horizon estimation (MHE)) designed to estimate the states and parameters for frequency dynamics of a power system. A simulation study was conducted using Matlab/Simulink by introducing Gaussian and non-Gaussian noise in the measurements. Results under Gaussian noise showed similar accuracy performance for all filters. EKF and UKF presented convergence or numerical instability issues due to incorrect initial guesses of parameters. MHE did not present convergence issues, however, required comparatively higher computation time. Nonetheless, the MHE could still be implemented in real-time for state and parameter estimation of power system. The impact of non-Gaussian noise on the methods was inconclusive and will require further study.