Enhancing the reliability of protection scheme for PV integrated microgrid by discriminating between array faults and symmetrical line faults using sparse auto encoder. Issue 2 (26th November 2018)

• Record Type: Journal Article
• Authors: Manohar, Murli; Koley, Ebha; Ghosh, Subhojit
• Is Part Of: IET renewable power generation. Volume 13:Issue 2(2019)
• Page Start: 308
• Page End: 317
• Publication Date: 2018-11-26
• Subjects: support vector machines -- learning (artificial intelligence) -- decision trees -- power generation protection -- power engineering computing -- neural nets -- photovoltaic power systems -- fault diagnosis -- power generation faults -- distributed power generation -- power grids -- unsupervised learning; PV integrated microgrid -- symmetrical line faults -- sparse auto encoder -- photovoltaic integrated microgrid -- reliable protection scheme -- similar voltage–current profile -- PV array faults -- conventional protection schemes -- sparse autoencoder -- fault detection -- SAE -- deep neural network approach -- array faults -- fault classification -- fault section identification -- grey‐scale image dataset -- unsupervised feature learning -- reliability analysis -- artificial neural network -- support vector machine -- decision tree‐based techniques -- grid‐connected mode -- islanding mode -- OPAL‐RT digital simulator; Renewable energy sources -- Periodicals; 333.79405
• DOI: 10.1049/iet-rpg.2018.5627
• Languages: English
• ISSNs: 1752-1416

Abstract

Abstract : The ever increasing power demand and stress on reducing carbon footprint have paved the way forwidespread use of photovoltaic (PV) integrated microgrid. However, thedevelopment of a reliable protection scheme for PV integrated microgrid ischallenging because of the similar voltage‐current profile of PV array faultsand symmetrical line faults. Conventional protection schemes based onpre‐defined threshold setting are not able to distinguish between PV array andsymmetrical faults, and hence fail to provide separate controlling actions forthe two cases. In this regard, a protection scheme based on sparse autoencoder(SAE) and deep neural network has been proposed to discriminate between arrayfaults and symmetrical line faults in addition to perform mode detection, faultdetection, classification and section identification. The voltage‐currentsignals retrieved from relaying buses are converted into grey‐scale images andfurther fed as input to the SAE to perform unsupervised feature learning. Theperformance of the proposed scheme has been evaluated through reliabilityanalysis and compared with artificial neural network, support vector machine anddecision tree based techniques under both islanding and grid‐connected mode ofthe microgrid. The scheme has been also validated for field applications byperforming real‐time simulations on OPAL‐RT digital simulator.