Comparison of Dynamic State Estimation Methods in the Time Domain

Correct information about a power system’s dynamic variables is important and necessary for protection and control issues. Today's power systems, which differ from past systems, face new challenges due to converter-based resources. A solution to these challenges is dynamic state estimation in short time intervals, such as the time domain. This paper simulates a standard 68-bus system in the presence of converter-based resources with a high penetration percentage in DIgSILENT software and compares the performance of four Bayesian filters in estimating the dynamic variables of the synchronous generators of the system using values in the time domain with each other. The four types of filters used include extended Kalman filter, unscented Kalman filter, ensemble Kalman filter, and particle filter.The MATLAB software suite was used for the comparison of the performance of the four filter types in different scenarios, including the presence of measurement and processing noise, extreme noise, network fault, data missing, state estimation time by each filter, and the comparison of time domain method with other methods such as phasor domain. Finally, the advantages and disadvantages of each were identified.