文摘
In this paper, a new brushless exciter generator power system stabilizer is proposed. The design is based on a recurrent neural network trained with a model-free approach and using the feed-forward error propagation learning algorithm. This is of great importance, as it will be outlined in the paper. The aim is to ensure a good damping of the power grid oscillations and to maintain constant voltage magnitude. This is done by providing an adequate control signal that delivers the reference input of the automatic voltage regulator. This stabilization signal is developed from the rotor speed. The results show the effectiveness of the proposed approach. The system response has less oscillations with a shorter transient time. The study was extended to faulty power plants.