Robust recurrent neural network-based dynamic equivalencing in power system

In this paper a new approach in dynamic equivalencing for power systems using robust recurrent artificial neural networks (ANN) as nonlinear dynamic equivalent is proposed as new alternative to the conventional way in dynamic equivalencing. The classical steps to generate dynamic equivalents are replaced by the robustly trained recurrent ANN taking into consideration a nearly global training process, in which the effect of the disturbance influence of the internal area on the external area has to be considered. The proposed approach is based on the nonlinear modeling and identification of dynamic systems for forming robust dynamic equivalents in large interconnected power systems which can be applied to transient stability studies. Simulation results demonstrate the effectiveness, high accuracy and robustness of this approach on different large multi-machine power systems with 2 to 8 boundary nodes between the internal and external area.