Determination of local transient stability control based on neural networks

Abstract

In order to apply neural networks to practical power systems, two feature reduction methods are presented in this paper. One introduces the generator coherent clustering and electrical distance to reduce the number of features, and the other adopts a linear neural network to perform feature extraction and trains it together with all the sub-neural networks. Simulation results of the Northwest power system in China indicate the feasibility of local transient stability control decisions in large-scale power systems based on neural networks.