Railway turnout support deterioration estimation under flooding condition using machine learning

Jessada Sresakoolchai, Sakdirat Kaewunruen, Mehmet Hamarat

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Downloads (Pure)

Abstract

Railway turnout is a critical component of the railway system. At the same time, it is a weak point in the railway structure because it is movable and contains connections. In the case of the ballast structure, railway turnouts are supported by crushed rocks which continuously deteriorate. However, the support may deteriorate rapidly in some situations such as flooding. Flooding can severely deteriorate the stability of the railway structure. Railway operations are critically affected by this deterioration because the maintenance will disturb regular operations. An ability to estimate the deterioration of the railway turnout support will enhance the efficiency of the maintenance plan. This study aims to develop a machine learning model to estimate the railway turnout support deterioration.
The estimation is conducted based on deterioration severity classification. The machine learning technique used to develop the model is a convolutional neural network. A key parameter used to estimate the railway turnout support deterioration is axle box acceleration. Validated numerical models are used to simulate the railway turnout behavior under flooding conditions. The numerical models are developed using the finite element method. An expected contribution of this study is the developed machine learning model can be used to estimate the deterioration of railway turnout support which will be beneficial to railway maintenance.
Original languageEnglish
Title of host publication21st Nordic Seminar on Railway Technology
PublisherTampere University, Finland
Publication statusPublished - 21 Jun 2022

Fingerprint

Dive into the research topics of 'Railway turnout support deterioration estimation under flooding condition using machine learning'. Together they form a unique fingerprint.

Cite this