Research On Neural-Network-Based Transient Stability Assessment

Feature selection and input dimension reduction are of paramount im-portance to transient stability assessment based on neural networks. The accuracy of stability classification is mainly defined by the separability of the input spaces. This paper summarized a system feature set for transient stability classification, several methods for analyzing the separability of input space of transient stability classification are discussed, Tabu search-ing technique is employed to select an effective set of features from a large initial features set. The classification test shows that the presented method works very well for feature selection. Fisher linear recognition is employed to cut down the training sample set, the computation burden of the ANN training is alleviated very much, so the convergence performance is improved. Zhang Wenchao (Research on Neural-Network-Based Transient Stability Assessment) Directed by Prof. Gu Xueping...