Robust Dynamic State Estimation for DFIG via the Generalized Maximum Correntropy Criterion Ensemble Kalman Filter

Accurate the dynamic state estimation (DSE) of complex power systems with doubly-fed induction generators (DFIG) is a pressing issue that requires resolution. The ensemble Kalman filter (EnKF) is a powerful tool for DFIG dynamic state estimation due to its superior tracking capabilities. However, the original EnKF with mean square error (MSE) cost may produce inaccurate results when measurement data is contaminated by non-Gaussian noise or outliers. To address this issue, we propose a novel robust EnKF method, denoted as GMCC-EnKF, which incorporates the generalized maximum correntropy criterion (GMCC) into the EnKF framework for DSE. The GMCC with higher-order moments of the error distribution is robust in non-Gaussian noise cases, making it an ideal replacement for MSE as the cost function of EnKF. Additionally, we use a statistical linear regression model and the fixed-point iteration method within the EnKF framework to solve the optimal state under the GMCC. Due to the potential influence of outliers on the innovation vector, we propose an enhanced GMCC-EnKF (EnGMCC-EnKF) by introducing an exponential function of the innovation vector to update the measurement noise covariance. We perform several numerical simulations using the improved IEEE 39 bus test system with DFIG, and the results demonstrate that the proposed method guarantees accuracy and maintains good robustness under measurement data with non-Gaussian noise (or outliers) cases.