An artificial neural network model as a preliminary track design tool

The formula derived from Zimmermann’s theory is commonly used in railway track design. However, this formula depends on variables such as the ballast coefficient, which are difficult to determine. In recent years, numerical models have been widely used as they allow the track to be studied as a comp...

Full description

Saved in:
Bibliographic Details
Published in:Proceedings of the Institution of Mechanical Engineers. Part F, Journal of rail and rapid transit Journal of rail and rapid transit, 2016-05, Vol.230 (4), p.1105-1117
Main Authors: Domingo, Laura Montalbán, Fernández-Villa, Juan Antonio Villaronte, Sendra, Claudio Masanet, Herráiz, Julia I. Real
Format: Article
Language:English
Subjects:
Artificial intelligence
Artificial neural networks
Computing time
Design engineering
Design optimization
Finite element method
Independent variables
Learning theory
Mathematical models
Neural networks
Numerical models
Rails
Railway engineering
Railway networks
Railway tracks
Tracking
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 formula derived from Zimmermann’s theory is commonly used in railway track design. However, this formula depends on variables such as the ballast coefficient, which are difficult to determine. In recent years, numerical models have been widely used as they allow the track to be studied as a complete system in which the input variables are known. However, the computation time of numerical models is often very large. This paper presents a pre-design tool that is based on an artificial neural network (ANN). This tool permits the efficient determination of the independent variables of the model, which depend on the track characteristics, the height of the embankment and the quality of the material used to form the embankment. The main advantage of the ANN model is the optimization of the design process, providing a pre-design scenario in which the independent variables are calculated on the basis of the vertical displacement of the rail top, which is the output of the ANN. This leads to significant savings in the computational time required to solve the finite element model.
ISSN:0954-4097
2041-3017
DOI:10.1177/0954409715576366