Legacy SE to distributed dynamic state estimators: Evolution and experience

This paper describes the evolution of state estimators from the legacy static state estimators to the distributed dynamic state estimators. Dynamic state estimators are computational demanding and only distributed approaches are practical. We present two specific distributed dynamic state estimators: (a) the quasi-dynamic state estimator which uses GPS-synchronized and non-synchronized measurements of phasors and a quasi-dynamic model (electrical transients are neglected but electromechanical transients are modeled), and (b) a distributed dynamic state estimator using sampled value measurements and the full dynamical model of the system. For each one of the distributed state estimators three solution approaches are used: (a) unconstraint optimization problem, (b) constraint optimization problem and (c) extended Kalman filter approach. The quasi-dynamic state estimator has been installed in pilot projects and field experience has been gained. The distributed dynamic state estimator forms the core technology for the setting-less protection and it has been extensively tested in the laboratory. Field demonstration are planned in the near future. Dynamic state estimator provide so much more and useful information that will drastically increase the operational reliability of the system. Recent developments have made dynamic state estimators feasible with practically no additional costs. By necessity, dynamic state estimators must be distributed operating on local data since we believe that it is impractical to construct massive and fast communication network for transferring the data that will be required by a centralized dynamic state estimator.