Maximum Correntropy Extended Kalman Filtering for Power System Dynamic State Estimation

Abstract

This paper discusses dynamic state estimation (DSE) of the multi-machine power systems in presence of non-Gaussian measurement noises. The extended Kalman filter (EKF) has been employed for performing DSE. Although EKF usually performs well under Gaussian noises, under heavy-tailed non-Gaussian measurement noises its performance may deteriorate. In order to improve the robustness of EKF, we have applied an EKF algorithm with maximum correntropy criterion (MCCEKF) to DSE. With the MCC integrated into EKF, higher-order statistics of data can thus be considered. The performance of the proposed algorithm is validated through simulation results on the WSCC 3-machine 9-bus system and NPCC 48-machine 140-bus system. For WSCC 3-machine system EKF and MCCEKF have similar performance under Gaussian measurement noises. However, under non-Gaussian (Cauchy, Laplace, Mixed) noises the performance of EKF deteriorates. For NPCC 48-machine system, EKF fails to converge whereas MCCEKF can still provide good estimation of the dynamic states under different measurement noises.