Neural network based for voltage stability prediction in power system / Muhamad Nasrul Hassan

This thesis presents the application of artificial Neural Network (ANN) to predict the voltage stability in power system network. The ANN type of Levenberg-Marquardt training back propagation is used in the voltage stability assessment. This requires a set of data to train the ANN in predicting the voltage stability accurately. A set of training data is generated based on the voltage stability index. After training the ANN then the other set of data is tested to ensure that the robustness of network is capable to predict the voltage stability accurately. The ANN output is coded into two categories that represents as the secure and not secure. The secure codes signify that the voltage condition of the system is stable with respect to the increased amount of reactive power at a particular load bus. On the other hand, the not secure codes signify that the voltage condition of the system is unstable with respect to the increased amount of reactive power at a particular load bus. Therefore, the output of ANN reflects the reactive power variations at a particular load bus. The effectiveness of ANN in predicting the voltage stability is analyzed with a case study of IEEE 9-bus reliability test system.