ANN-based robust DC fault protection algorithm for MMC high-voltage direct current grids

Wang Xiang, Saizhao Yang, Jinyu Wen

Research output: Contribution to journalArticle

Abstract

Fast and reliable protection is a significant technical challenge in modular multilevel converter (MMC) based DC grids. The existing fault detection methods suffer from the difficulty in setting protective thresholds, incomplete function, insensitivity to high resistance faults and vulnerable to noise. This paper proposes an artificial neural network (ANN) based method to enable DC bus protection and DC line protection for DC grids. The transient characteristics of DC voltages are analysed during DC faults. Based on the analysis, the discrete wavelet transform (DWT) is used as an extractor of distinctive features at the input of the ANN. Both frequency-domain and time-domain components are selected as input vectors. A large number of offline data considering the impact of noise is employed to train the ANN. The outputs of the ANN are used to trigger the DC line and DC bus protections and select the faulted poles. The proposed method is tested in a four-terminal MMC based DC grid under PSCAD/EMTDC. The simulation results verify the effectiveness of the proposed method in fault identification and the selection of the faulty pole. The intelligent algorithm based protection scheme has good performance concerning selectivity, reliability, robustness to noise and fast action.
Original languageEnglish
JournalIET Renewable Power Generation
Early online date1 Nov 2019
DOIs
Publication statusE-pub ahead of print - 1 Nov 2019

Fingerprint

Neural networks
Electric potential
Poles
Discrete wavelet transforms
Fault detection

Keywords

  • modular multilevel converter
  • DC grids
  • discrete wavelet transform (DWT)
  • protection schemes

Cite this

@article{b05c02d8554f417d90d7fc7699d0f6c8,
title = "ANN-based robust DC fault protection algorithm for MMC high-voltage direct current grids",
abstract = "Fast and reliable protection is a significant technical challenge in modular multilevel converter (MMC) based DC grids. The existing fault detection methods suffer from the difficulty in setting protective thresholds, incomplete function, insensitivity to high resistance faults and vulnerable to noise. This paper proposes an artificial neural network (ANN) based method to enable DC bus protection and DC line protection for DC grids. The transient characteristics of DC voltages are analysed during DC faults. Based on the analysis, the discrete wavelet transform (DWT) is used as an extractor of distinctive features at the input of the ANN. Both frequency-domain and time-domain components are selected as input vectors. A large number of offline data considering the impact of noise is employed to train the ANN. The outputs of the ANN are used to trigger the DC line and DC bus protections and select the faulted poles. The proposed method is tested in a four-terminal MMC based DC grid under PSCAD/EMTDC. The simulation results verify the effectiveness of the proposed method in fault identification and the selection of the faulty pole. The intelligent algorithm based protection scheme has good performance concerning selectivity, reliability, robustness to noise and fast action.",
keywords = "modular multilevel converter, DC grids, discrete wavelet transform (DWT), protection schemes",
author = "Wang Xiang and Saizhao Yang and Jinyu Wen",
year = "2019",
month = "11",
day = "1",
doi = "10.1049/iet-rpg.2019.0733",
language = "English",
journal = "IET Renewable Power Generation",
issn = "1752-1416",
publisher = "Institution of Engineering and Technology",

}

ANN-based robust DC fault protection algorithm for MMC high-voltage direct current grids. / Xiang, Wang; Yang, Saizhao; Wen, Jinyu.

In: IET Renewable Power Generation, 01.11.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - ANN-based robust DC fault protection algorithm for MMC high-voltage direct current grids

AU - Xiang, Wang

AU - Yang, Saizhao

AU - Wen, Jinyu

PY - 2019/11/1

Y1 - 2019/11/1

N2 - Fast and reliable protection is a significant technical challenge in modular multilevel converter (MMC) based DC grids. The existing fault detection methods suffer from the difficulty in setting protective thresholds, incomplete function, insensitivity to high resistance faults and vulnerable to noise. This paper proposes an artificial neural network (ANN) based method to enable DC bus protection and DC line protection for DC grids. The transient characteristics of DC voltages are analysed during DC faults. Based on the analysis, the discrete wavelet transform (DWT) is used as an extractor of distinctive features at the input of the ANN. Both frequency-domain and time-domain components are selected as input vectors. A large number of offline data considering the impact of noise is employed to train the ANN. The outputs of the ANN are used to trigger the DC line and DC bus protections and select the faulted poles. The proposed method is tested in a four-terminal MMC based DC grid under PSCAD/EMTDC. The simulation results verify the effectiveness of the proposed method in fault identification and the selection of the faulty pole. The intelligent algorithm based protection scheme has good performance concerning selectivity, reliability, robustness to noise and fast action.

AB - Fast and reliable protection is a significant technical challenge in modular multilevel converter (MMC) based DC grids. The existing fault detection methods suffer from the difficulty in setting protective thresholds, incomplete function, insensitivity to high resistance faults and vulnerable to noise. This paper proposes an artificial neural network (ANN) based method to enable DC bus protection and DC line protection for DC grids. The transient characteristics of DC voltages are analysed during DC faults. Based on the analysis, the discrete wavelet transform (DWT) is used as an extractor of distinctive features at the input of the ANN. Both frequency-domain and time-domain components are selected as input vectors. A large number of offline data considering the impact of noise is employed to train the ANN. The outputs of the ANN are used to trigger the DC line and DC bus protections and select the faulted poles. The proposed method is tested in a four-terminal MMC based DC grid under PSCAD/EMTDC. The simulation results verify the effectiveness of the proposed method in fault identification and the selection of the faulty pole. The intelligent algorithm based protection scheme has good performance concerning selectivity, reliability, robustness to noise and fast action.

KW - modular multilevel converter

KW - DC grids

KW - discrete wavelet transform (DWT)

KW - protection schemes

UR - https://digital-library.theiet.org/content/journals/iet-rpg

U2 - 10.1049/iet-rpg.2019.0733

DO - 10.1049/iet-rpg.2019.0733

M3 - Article

JO - IET Renewable Power Generation

JF - IET Renewable Power Generation

SN - 1752-1416

ER -