An Active Power Failure Early Warning Probability Model Based on Support Vector Machine Algorithm

Abstract

The occurrence of power grid blackout will cause major production equipment damage, cause major economic losses, or even cause casualties in some extremely serious cases. It is necessary to upgrade the technology of traditional power grid so that emergency measures can be taken when unexpected cascade failure occurs. In this work, an active power failure warning probability model is proposed for smart grid warning system. First, probabilistic evaluation of system performance is carried out to generate the historical database required by cascade state. Then, the support vector machine (SVM) algorithm is used to train the historical database, and power failure is predicted by the support vector machine (SVM) binary classifier. Finally, the proposed model is validated by IEEE 30 node system. The proposed model can well understand the essence of cascade failure and can proactively predict cascade failure, which is helpful for the planning and maintenance of power grid system.