THE MODEL OF REMOTE MONITORING SYSTEM FOR ELECTRIC ROLLING STOCK CONDITION USING ARTIFICIAL NEURAL NETWORKS
Rubrics: ARTICLES
Authors:
1.
Emperor Alexander I St. Petersburg State Transport University
2.
Emperor Alexander I St. Petersburg State Transport University
3.
Emperor Alexander I St. Petersburg State Transport University
Type:
Article
Pages:
from 27
to 37
Status:
Published
Received:
11.01.2025
Accepted:
11.01.2025
Published:
11.01.2025
Subject area:
UDK УДК
Language:
Russian
Keywords:
artificial neural networks, neural network technologies, information technologies, railway transport, remote monitoring
Abstract (English):
The paper analyzes tools for automated monitoring of rolling stock, establishes requirements, and develops a model of information system to monitor the characteristics of the rolling stock. A number of artificial neural network architectures have been analyzed, and a convolutional neural network algorithm for determining the percentage of the rolling stock current collector wear on the basis of pictures has been proposed. The algorithm includes image preprocessing, creation of convolutional neural network model, its training and use for classification of new images.
The paper analyzes tools for automated monitoring of rolling stock, establishes requirements, and develops a model of information system to monitor the characteristics of the rolling stock. A number of artificial neural network architectures have been analyzed, and a convolutional neural network algorithm for determining the percentage of the rolling stock current collector wear on the basis of pictures has been proposed. The algorithm includes image preprocessing, creation of convolutional neural network model, its training and use for classification of new images.
Keywords:
artificial neural networks, neural network technologies, information technologies, railway transport, remote monitoring
artificial neural networks, neural network technologies, information technologies, railway transport, remote monitoring
1.
