Multi-class support vector machines for classification of transmission line faults

Abstract

This paper presents an approach based on multi-class support vector machine (MSVM) and wavelet transform (WT) for classifying high-voltage transmission line faults. The proposed method uses one terminal current and voltage information obtained from a 750 kV power transmission line model. Before the training of the support vector machines, WT is employed for feature extraction. Wavelet entropy criterion is applied to wavelet detail coefficients to reduce feature vector in size. Different fault conditions and locations are considered and it has been shown that the proposed method yields very satisfactory results with 2.77% average error.