A Hybrid Artificial Neural Network for Voltage Security Evaluation in a Power System

A majority of recent large-scale blackouts have been the consequence of instabilities characterized by sudden voltage collapse phenomena. This paper presents a method for voltage instability monitoring in a power system with a hybrid artificial neural network which consist of a multilayer perceptron and the Kohonen neural network. The proposed method has a couple of the following functions: the Kohonen network is used to classify the system operating state; the Kohonen output patterns are used as inputs to train of a multilayer perceptron for identification of alarm states that are dangerous for the system security. The approach is targeting a blackout prevention scheme; given that the blackout signal is captured before it can collapse the power system. The proposed method is realized in R and demonstrated the modified IEEE One Area RTS-96 power system.