DEVELOPMENT OF DSP-ALGORITHM FOR DETECTION OF FAULTS ON TRANSMISSION LINES

Abstract

Transmission line is one of the most essential elements in the power system functions as it connects generation and loads. Faults on transmission lines not only affect the equipment but also the power quality. Disturbances (Faults) which affect the transmission line system should be detected accurately and promptly. Fault detection and classification on transmission lines is imperative for economic operation of power system components as it facilitates quicker repair, improve system availability, reduce operating costs, and saves time and space. In the present work the automated fault data generation model is done by using PSCAD/EMTDC. To illustrate the effectiveness of both the methods which are used for fault classification, extensive simulations have been carried out for different fault conditions with wide variations in the operating conditions such as fault inception angles and fault resistances. In the present work, Fault classification on three phase transmission lines using Artificial Neural Networks (ANN) and Support Vector Machine are described and the results obtained by both the methods have been compared. Wavelet transform is used for the decomposition of measured three line current signals and for extraction of the most significant features (feature extraction), which facilitates the training of both ANN and SVM classifiers. The ground detection is carried out by proposed ground index in the case of ANN classification. Although the accuracy/efficiency obtained by ANN classification is quite acceptable, it suffers with some deficiencies. So, SVMs have been used to overcome these deficiencies, particularly in terms of getting better classification performance (high accuracy). Only one SVM classifier is used for obtaining a decision of a fault or no fault on any phase or multiple phases of a Transmission line. The Kernel function used for SVM classifier in this report is radial basis function (RBF) since it provides better fault classification accuracy. Key words: Three phase two terminal transmission line, wavelet transform, Artificial Neural Networks (ANNs), Support Vector Machine (SVM), feature extraction, fault classification accuracy/efficiency.