Power System Transient Stability Assessment Based on Adaboost and Support Vector Machines

Power system stability is an important problem for secure system operation. Transient stability is one of key problems of power system stability. In this paper, support vector machines (SVM), a novel type learning method and based on statistical learning theory, is applied to assess the transient stability of power system after faults occur on transmission lines. Reactive and active powers of all generators after fault cleaning and abstract attributes consisted of the inputs of SVM. An effective feature selection technique was used to refine the inputs and increase the accuracy of SVM. As the comparison, we considered three kinds of classifiers, namely ANN, Decision tree and the SVM. Our results showed that the SVM with RBF kernel achieved the highest classification accuracy, followed by the SVC with polynomial kernel on IEEE 16-generator and 50-generator test systems, in transient stability analysis, SVM classifier showed superiority over traditional methods.