An Effective Zone-3 Supervision of Distance Relay for Enhancing Wide Area Back-Up Protection of Transmission System

• Published in: IEEE transactions on power delivery Vol. 36; no. 5; pp. 3204 - 3213
• Main Authors: Sahoo, Biswajit, Samantaray, Subhransu Ranjan, Bhalja, Bhavesh R.
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.10.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Artificial neural networks; Data mining; Deep Neural Network; Distance Relay; Generators; Information retrieval; Load Encroachment; Machine learning; Measuring instruments; Out-of-step; Phasor Measurement Unit; Phasor measurement units; Phasors; Power Swing; Power system dynamics; Power system stability; Protective relaying; Real time; Real Time Digital Simulator; Real-time systems; Relay; Relaying; Switches; Wide Are Backup protection
• ISSN: 0885-8977; 1937-4208
• DOI: 10.1109/TPWRD.2020.3035885

Distance relays are quite vulnerable to maloperation during dynamic stressed situations. The apparent impedance seen by the relay encroaches the zone-3 resulting in distance relay maloperation that can initiate cascading failures. This paper presents an enhanced scheme for wide area backup protection which implements improved protection algorithms to decide final relaying status. A machine learning tool namely Deep Neural Network is utilized to build those protective relaying decision logics: RDL-1 and RDL-2. Wide area information retrieved from phasor measurement units (PMU) is used to take real time decisions. RDL-1 is designed to detect if the system is operating in normal or stressed condition. During normal state, traditional zone-3 relay undertakes the relaying decision. However, under stressed situations, RDL-1 switches the decision-taking to RDL-2. RDL-1 also predicts whether the system is progressing towards an out-of-step situation so that appropriate remedial action can be undertaken. Proposed scheme was studied on IEEE-39 bus New England system using MATLAB/SIMULINK and validated on Real Time Digital Simulator platform and third zone mal-tripping are avoided. Performance of the proposed algorithm was compared with the conventional blinder and other data-mining techniques which show considerably high classification accuracy.