Adaptive Robust Unscented Kalman Filter for Power System Dynamic State Estimation

This paper proposes a new singular value decomposition-based adaptive robust unscented Kalman filter (SVD-ARUKF) for power system dynamic state estimation. Specifically, a suboptimal fading factor is introduced into the predictive covariance matrix to improve the tracking performance for sudden state variables. Unknown system noise is adapted through the covariance matching method. To avoid the numerical sensitivity, SVD is performed on the covariance matrix. Simulated results demonstrate the filtering stability and accuracy of the proposed SVD-ARUKF algorithm on the IEEE 30-node system.