Dynamic State Estimation of Synchronous Machines Using Robust Cubature Kalman Filter Against Complex Measurement Noise Statistics

TL;DR

This paper introduces a robust Cubature Kalman Filter (RCKF) that combines Huber's M-estimation with classical CKF to improve dynamic state estimation of synchronous machines under complex, non-Gaussian noise and bad data conditions.

Contribution

The paper develops a new RCKF algorithm that enhances CKF's robustness against non-Gaussian noise and bad data by integrating Huber's M-estimation, improving estimation accuracy and convergence.

Findings

RCKF outperforms classical CKF in accuracy and convergence.

RCKF effectively mitigates the impact of bad data.

Simulation results on IEEE systems validate the method's effectiveness.

Abstract

Cubature Kalman Filter (CKF) has good performance when handling nonlinear dynamic state estimations. However, it cannot work well in non-Gaussian noise and bad data environment due to the lack of auto-adaptive ability to measure noise statistics on line. In order to address the problem of behavioral decline and divergence when measure noise statistics deviate prior noise statistics, a new robust CKF (RCKF) algorithm is developed by combining the Huber's M-estimation theory with the classical CKF, and thereby it is proposed to coping with the dynamic state estimation of synchronous generators in this study. The simulation results on the IEEE-9 bus system and New England 16-machine-68-bus system demonstrate that the estimation accuracy and convergence of the proposed RCKF are superior to those of the classical CKF under complex measurement noise environments including different measurement noises and bad data, and that the RCKF is capable of effectively eliminating the impact of bad data on the estimation effects.