Data-based Generation Rescheduling to Mitigate poorly damped Oscillations in Transmission Systems

In this paper we present a purely data-based algorithm for power system oscillation damping using generation rescheduling. The algorithm builds a database over historically observed operating states with known sufficient damping. In case an operating state with poor damping is observed, the algorithm searches through the database to find the closest operating state. To determine the closeness between two operating states we test three different features based on measurements of active power. The algorithm can provide power system operators with a suggestion on which generators to reschedule to move the system state towards a state with sufficient damping. The algorithm is tested on an aggregated power system model of the Nordic power system using historic operating state data from the NordPool market data. We demonstrate the value of using the algorithm for preventive rescheduling. Moreover, we demonstrate how the algorithm performs on several operating states from the first two weeks of January 2019.