Deep Neural Network-based Wide Area Back-up Protection Scheme for Transmission System

Abstract

During dynamic stressed conditions, distance relays are utterly vulnerable to mal-operations. The apparent impedance seen by the relay can enter into the third zone causing distance relay mal-tripping which may lead to cascading outages. This work proposes a new methodology for wide-area backup protection that incorporates two different decision logics for final relaying decision. Phasor measurement unit (PMU) data is utilized to determine whether the system is in normal or stressed state. A Deep learning based method known as Deep Neural Network (DNN) is implemented to develop first decision logic i.e. state assessment. During normal condition, conventional distance relay takes the relaying decision, but during stressed condition the decision switches to the second protective relaying logic which is based on wide area information. Proposed scheme was tested for WSCC 9 bus system using MATLAB/SIMULINK platform and mal-operations of zone-3 distance relay are prevented. Performance comparison of the proposed logic with the blinder-based techniques exhibited substantially higher blocking efficiency.