An Adaptive Distance Protection Scheme for High Varying Fault Resistances

Abstract

The variation of fault resistances introduces error in the measured apparent impedance of the conventional distance protection scheme. With this, the measured apparent impedance at the relay location is not proportional to its length. This paper presents an adaptive protection scheme using Artificial Neural Network (ANN) to address the problem. A MATLAB based adaptive distance relaying scheme is proposed using the ANN feed-forward nonlinear supervised back-propagation algorithm based on the Levenberg-Marquardt network topology. The PSCAD /EMTDC software is used to generate the current and voltage signals for specified transmission lines which are used for subsequent ANN training and testing of the proposed algorithm. The proposed non-conventional adaptive scheme was validated with a new set of high fault resistances data using the proposed model. The results show the ability of ANN to correctly detect, classify and localized fault under varying fault resistance.