An Improved Cubature Kalman Filter for State Estimation Against Measurement Outliers

Accurate state estimation is critical for power grids, significantly contributing to their safety, stability, and reliability. However, the presence of channel noise and outliers stemming from phasor measurement units renders the noise model a deviation from the Gaussian distribution. To mitigate this challenge, this paper introduces a parameterized analytical update cubature Kalman filter (PACKF) that significantly enhances estimation accuracy. Firstly, the updated analytical form of the state variable is derived, in which unknown parameter is introduced. Secondly, the unknown parameter is approximated using fixed-point iteration, followed by the analytical computation of the required joint posterior PDF. Finally, extensive simulations indicatie that the PACKF commendable accuracy and efficiency across diverse scenarios.