Power System Dynamic Security Assessment through a Neuro-Fuzzy Scheme

This paper proposes a neuro-fuzzy based methodology for power system dynamic security assessment, focusing on transient stability. The proposed method utilizes a concurrent neuro-fuzzy system based on dynamic performance indices calculation. An Artificial Neural Network (ANN) is trained to determine the performance indices, which allow the classification of the operative system state security. The ANN inputs are composed by signals that indicate the disturbance severity and its ability to reestablish an acceptable equilibrium point. A diagnosis fuzzy is applied to the ANN result in order to classify the security state and to indicate the necessity of detailed analysis in doubtful situations. In this way, only doubtful classifications will require full time-domain simulations, which is a very time consuming task. The proposed methodology was applied to a test system that represents part of the South region of the Brazilian electric system. The results have shown the potential benefits of the proposed scheme.