Modern Approaches For Protection Of Autoreclosing Ehv Transmission Line

Abstract EN

This paper proposes two approaches based on Support Vector Machine (SVM) and Extreme Learning Machine (ELM) for discrimination between line charging inrush current and the current resulting from auto- reclosing a faulted EHV transmission line.

The wavelet transform is first applied to decompose the current samples into a series of wavelet coefficients.

These wavelet coefficients are then employed to train SVM and ELM to classify transient events in auto-reclosing EHV transmission line.

The merit of this method is demonstrated by simulation of different faults and switching conditions in a transmission line using MATLAB Simulink.

The proposed methods using SVM and ELM are tested on a 230-kV, 144.4-km transmission line by varying fault resistance, fault distance, fault inception angle and switching angle.

The performance of SVM and ELM is compared in terms of training time and classification accuracy with Probabilistic Neural Network (PNN) and Back propagation Neural Network (BPNN).

The results proved that the classification accuracy of SVM is better than ELM, PNN and BPNN with less number of training samples.

Also, SVM is robust to parameter variations such as fault resistance, fault inception angle and fault distance.