Application of Kalman filter based estimation techniques to electric power systems

Abstract

This thesis presents several applications of dynamic state estimators based on Kalman filtering to different fields of the electric power systems. First, a parameter estimation technique is proposed, applied to a generation set composed by the synchronous machine along with the frequency regulation (speed governor) and the voltage controllers (automatic voltage regulator and power system stabilizer). The proposed method is based on a formulation of the unscented Kalman filter, being this study the first attempt, to the authors’ knowledge, to include the full generation set in the estimator model, with the corresponding state variables and parameters, using just external measurements taken at the generator terminal bus. A similar estimation technique, using the cubature Kalman filter, is implemented subsequently for a joint estimation of the dynamic state and the model parameters of a variable speed wind turbine with permanent magnet synchronous generator and back to back voltage source converter. In this case, the major contribution consists of the inclusion of the control parameters in the state vector to be estimated. Finally, three Kalman filter formulations (unscented Kalman filter, cubature Kalman filter and ensemble Kalman filter) are implemented to address the problem of identifying the electrical phase of single phase consumers in distribution grids, using for this purpose hourly energy measurements exclusively. The accuracy and robustness of these estimators are compared in different case studies with variations in the number of loads and errors in the measurements and the considered model.