Swetapadma, Aleena, Agarwal, Shobha, Abdelaziz, Almoataz Y., Kotb, Hossam, AboRas, Kareem M., Flah, Aymen and Shouran, Mokhtar ![]() ![]() |
![]() |
PDF
- Published Version
Available under License Creative Commons Attribution. Download (1MB) |
Abstract
Discrimination of different DC faults near a converter end of a DC section consisting of a filter, a smoothing reactor, and a transmission line is not an easy task. The faults occurring in the AC section can be easily distinguished, but the internal and near-side external faults in the DC section are very similar, and the relay may cause false tripping. This work proposes a method to distinguish external and internal faults occurring in the DC section. The inputs are the voltage signals at the start of the transmission line and the end of the converter filter. The difference in voltage signals is calculated and given to an intelligent controller to detect and discriminate the faults. The intelligent controller is designed using machine learning (ML) and deep learning (DL) techniques for fault detection. The long short-term memory (LSTM-) based relay gives better results than other ML methods. The proposed method can distinguish internal from external faults with 100% accuracy. Another advantage is that a primary relay is suggested that detects faults quickly within a fraction of milliseconds. Nevertheless, another advantage is that a backup relay has been designed in case the primary relay cannot operate. Results show that the LSTM-based protection scheme provides higher sensitivity and reliability under different operation modes than the conventional traveling wave-based relay.
Item Type: | Article |
---|---|
Date Type: | Publication |
Status: | Published |
Schools: | Engineering |
Additional Information: | License information from Publisher: LICENSE 1: URL: http://creativecommons.org/licenses/by/4.0/ |
Publisher: | Frontiers Media |
Date of First Compliant Deposit: | 6 October 2022 |
Date of Acceptance: | 1 September 2022 |
Last Modified: | 10 Feb 2024 02:10 |
URI: | https://orca.cardiff.ac.uk/id/eprint/153118 |
Actions (repository staff only)
![]() |
Edit Item |