A UKF-based approach to estimate parameters of a three-phase synchronous generator model

This paper proposes an approach based on the Unscented Kalman Filter to estimate the parameters of a three-phase synchronous generator model. The developed approach makes use of the trajectory sensitivity functions to assess the influence of the parameters in the model outputs and to classify the parameters into subgroups. Some of the parameters can only be estimated with the selection of a correct sampling window for the estimation process. This characteristic is taken into account to choose a set of windows for the application of the Unscented Kalman Filter. The covariance matrices of this filter, which are also important for the correct estimation of the parameters, are altered to obtain better results. The assessment of the impact of parameter initial values in the filter is also presented in this work. The signals used in the proposed approach are measured at the terminal bus of the synchronous generator, which eliminates the need of knowing the parameters of the grid. The accuracy of the approach is evaluated by comparing the mean squared error between the measured quantities and the responses of the model.