Feature Selection for Small-Signal Stability Assessment
This paper introduces different feature selection techniques for neural network based small-signal stability assessment. Large-scale power systems like the European interconnected network may experience low frequency oscillations between remote parts of the system. These oscillations are caused by large power transits in the network. In dynamic security assessment, a fast and accurate artificial intelligence technique can be applied. Hereby, the state of the system is predicted by the use of a neural network (NN), which provides information about the system eigenvalues and therefore the damping of the oscillations. Because NN cannot be trained with the complete power system data, a reduction technique needs to be implemented. Therefore, this paper introduces different feature selection techniques and their applications.