“…In addition, the robustness of data-driven approaches to system and topology changes as well as missing or inaccuracies in system information has made them the focus of many researchers recently [1][2][3][4][5][6]. Examples of data-driven and machine-learning-based state estimation techniques in power systems include techniques based on neural networks [4,15], the Bayesian approach [3,16], minimum mean squared error (MMSE) [5,17], and the Kalman filter (KF) [2,18]. Since KF and its variations, including the extended Kalman filter (EKF) and the unscented Kalman filter (UKF), have a great ability to model and express dynamic systems, many dynamic state estimation algorithms for power systems are developed based on these techniques [2,14,19,20].…”