A robust unified data-driven protection scheme for hybrid AC/DC transmission lines

Abstract

To meet rising energy demands, reduce carbon emissions, and address climate change, the importance of hybrid AC/DC transmission has increased. The effective utilization of high-voltage DC in existing AC transmission lines will reduce costs and losses and increase the existing transmission capacity. Therefore, the protection schemes of fault detection, location, and classification are complicated because of the different characteristics of AC and DC in the same transmission network. The existing protection schemes mostly utilize threshold-based criteria and cannot ensure hybrid network protection. To overcome these challenges, a unified data-driven protection scheme (UDDPS) using distinctive features is proposed to detect and identify (either AC or DC) and classify both AC/DC faults of hybrid transmission lines (HTLs). The UDDPS has ensured the protection of AC/DC lines by timely intervening faults and effectively isolating the respective relays of the faulty lines. The fault current data are retrieved during a fault abnormality using distinctive retrieved features from the single end of the HTL. The proposed scheme applies to multi-transmission networks and is independent of the communication channels. Extensive simulation scenarios, such as fault/no-fault, varying fault locations, and near and far locations with low/high impedances, were examined in MATLAB/Simulink to train the UDDPS. The UDDPS successfully detected and identified hybrid AC/DC line faults with 100% accuracy and further classified the AC and DC faults with 97.61 % and 99.20 % accuracy. The efficacy and robustness of the proposed technique were confirmed by comparison with the existing schemes of support vector machine, decision tree, and LSTM. The UDDPS surpassed in terms of the overall model under a normal and noisy environment.