Fault Diagnosis of Tower Grounding Conductor Based on the Electromagnetic Measurement and Neural Network

The performance of the tower grounding conductor is very important for the safe and reliable operation of the power transmission system. Under the current excitation, the grounding conductor in different states will produce different magnetic fields on the earth's surface. By measuring the surf...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on instrumentation and measurement 2022, Vol.71, p.1-9
Main Authors: Lu, Caijiang, Zhang, Tianyi, Sun, Shaoheng, Cao, Zhongqing, Xin, Mingyong, Fu, Guoqiang, Wang, Tao, Wang, Xi
Format: Article
Language:English
Subjects:
Artificial neural networks
Computational modeling
Conductors
Corrosion
Deep neural network
Earth surface
Electrocutions
Electromagnetic measurement
Fault diagnosis
fine-tuning
Grounding
grounding conductor
Machine learning
Magnetic fields
Neural networks
Poles and towers
Task analysis
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The performance of the tower grounding conductor is very important for the safe and reliable operation of the power transmission system. Under the current excitation, the grounding conductor in different states will produce different magnetic fields on the earth's surface. By measuring the surface magnetic fields, the fault types and location of the grounding conductor can be evaluated. However, the process of traditional manual analysis is extremely complex. This article proposes a fault diagnosis method for tower grounding conductors based on deep learning to replace the manual diagnosis. The earth's surface magnetic field dataset generated by the grounding conductor consists of simulation data and experimental data. For practical application, the fine-tuning method is proposed to improve the diagnosis performance of the 1-D-convolutional neural network (1-D-CNN). Compared with the original 1-D-CNN model, the results demonstrate that the proposed fine-tuning 1-D-CNN has improved the diagnostic ability of the tower grounding conductor. For the six kinds of faults of rectangular tower grounding conductor, the average classification accuracy of the fine-tuning 1-D-CNN model reaches 80.50% when only the classifier and dense connection layers are trained. The proposed method can be used in the fault diagnosis of tower grounding conductors, which has significant potential applications in detecting and maintaining tower grounding conductors.
ISSN:0018-9456
1557-9662
DOI:10.1109/TIM.2022.3192070