Implementation of transient stability assessment using artificial neural networks

This paper aims at verifying the accuracy of Artificial Neural Networks (ANN) in assessing the transient stability of a single machine infinite bus system. The fault critical clearing time obtained through ANN is compared to the results of the conventional equal area criterion method. The multilayer feedforward artificial neural network concept is applied to the system. The training of the ANN is achieved through the supervised learning; and the back propagation technique is used as a learning method in order to minimize the training error. The training data set is generated using two steps process. First, the equal area criterion is used to determine the critical angle. After that the swing equation is solved using the point-to-point method up to the critical angle to determine the critical clearing time. Then the stability of the system is verified. As a result we find that the critical clearing time is predicted with slightly less accuracy using ANN compared to the conventional methods for the same input data sets unless the ANN is well trained.