Accurate, simultaneous and Real-Time screening of N-1, N-k, and N-1-1 contingencies

Abstract

Generally, in contingency analysis, N-1 events are studied. However, simultaneous failure of more than one device is more dangerous and might result in cascading failures. Due to cyber-attacks and malicious motivations, attacking several components of the system is more probable. Various methods have been proposed for contingency screening, which are mainly time-consuming and cannot be applied to real systems. On the other hand, there is no integrated algorithm for screening N-1, N-k, and N-1-1 contingencies. In this paper, an indirect method has been presented for this purpose. The proposed method extracts different critical contingencies through screening contingencies in offline mode for some selected operating modes. Then, they are clustered using the Gaussian-mixture model using the expectation–maximization algorithm. Therefore, the boundary between data is specified, which helps to find the closest (the most similar) operating modes. In online mode, the critical contingencies of the closest trained data are introduced as the critical contingency set of the present operating point. To the best of our knowledge, this study employs the correlation between the loads and generators in contingency screening for the first time. The proposed method outperforms the existing methods in terms of simplicity, accuracy, and speed such that it detects contingencies with the maximum accuracy of 100% without imposing any additional computational load.