On-line power systems voltage stability monitoring using artificial neural networks

A method for on-line voltage stability monitoring of a power system based on Multilayer Perceptron (MLP) neural network is proposed in this paper. Considering that the power system is operating under quasistatic conditions, by using power flow model and singular value decomposition of the reduced Jacobian matrix, a suitable index to quantify the proximity of power system voltage instability is defined. Then, a neuronal network is trained to learn the correlation between the key factors of the voltage stability phenomena and this index. Once trained, the neural network provides the above mentioned voltage stability index as output for a predefined set of input variables that are known as directly influencing the stability conditions of the power system. Since the input variables for the neural network may be obtained from the steady state estimator, the proposed method can be implemented as a function of the Energy Management System (EMS) for on-line voltage stability monitoring. Tests are carried out using the IEEE 30-bus system, where different operating scenarios are considered.