Faulty Line Identification in AC-DC Hybrid Grids Based on MTF and Improved Resnet

As the degree of AC-DC hybridization of power grids is increasing, their fault characteristics become more complex, and the current hybrid grid fault diagnosis methods, despite their high accuracy, are not sufficiently adaptable, so there is an urgent need to study new grid fault diagnosis methods....

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2023, Vol.11, p.119722-119732
Main Authors: Wu, Hao, Chen, Weizhe, Qi, Ziyuan, Song, Hong, Tian, Haipeng
Format: Article
Language:English
Subjects:
AC > DC hybrid grid
Algorithms
Artificial neural networks
Convolutional neural networks
Correlation coefficient
Data loss
data reconstruction
Deep learning
Fault diagnosis
fault line selection
Feature extraction
Hybrid power systems
Identification methods
Machine learning
MTF
Neural networks
Resnet
Transient analysis
Voltage
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:As the degree of AC-DC hybridization of power grids is increasing, their fault characteristics become more complex, and the current hybrid grid fault diagnosis methods, despite their high accuracy, are not sufficiently adaptable, so there is an urgent need to study new grid fault diagnosis methods. To this end, the Markov Transition Field combined with an improved Resnet fault line identification method for AC-DC hybrid grids is proposed. First, the data is reconstructed by the improved complete ensemble empirical mode decomposition with adaptive noise, and then the MTF is used to transform the one-dimensional signal into a two-dimensional picture, then on the basis of residual neural network, the original network is improved by adding multi-branch cavity convolution structure and Ghost module, and the fault features are adaptively extracted and classified by the improved network, so as to realize faulty line identification. The experimental results show that the proposed method can effectively identify the fault lines of AC-DC hybrid power grid, the improved residual neural network can dig out the fault features more deeply, and has strong anti-noise and anti-data loss interference ability, the method has 99.91% fault line identification accuracy. It has higher recognition performance compared to traditional machine learning algorithms and various deep learning algorithms.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2023.3327449