Koopman mode analysis of power systems oscillations

This paper deals with the application of a mathematical technique named Koopman Mode Analysis in electrical power systems to decompose its swing dynamics into modes of oscillations. A practical data-driven algorithm of the Koopman Mode Analysis is proposed to extract frequencies growth rates and norms of identified modes. The computed Koopman modes are ranked based on their growth rates. The Koopman analysis is applied to a two-area four-machine power system. The choice of the number of KM to be computed is discussed. A comparison of Koopman modes is carried out with linear modes identified by the conventional small-signal stability analysis. This comparison reveals the less damped dynamics which may or not be the linear modes. Such information is particularly useful for power system control in post severe disturbances.