Robust Power System Forecasting-Aided State Estimation With Generalized Maximum Mixture Correntropy Unscented Kalman Filter

Abstract

As an effective method for power system forecasting-aided state estimation (FASE), the unscented Kalman filter (UKF) based on correntropy has been widely used in recent years, ensuring the safe and reliable operation of power systems. In this article, to address the impulsive noise, Laplacian noise, bad measurement data, and sudden load change, a robust UKF algorithm based on generalized maximum mixture correntropy (GMMC-UKF) criterion is proposed for FASE, in which the kernel is composed of two generalized Gaussian functions. Specifically, we use a statistical linearization technique to unify the state error and measurement error in the cost function and obtain the optimal value of state estimation by fixed-point iteration. The effectiveness of the proposed algorithm for FASE is verified on IEEE 14-, 30-, and 57-bus test systems under a variety of abnormal situations. Compared with traditional correntropy algorithms, the GMMC-UKF shows more accurate estimation and stronger robustness.