Culwick, Richard (2023). Intelligent real-time monitoring of critical rail infrastructure. University of Birmingham. Ph.D.
|
Culwick2023PhD.pdf
Text - Accepted Version Available under License All rights reserved. Download (3MB) | Preview |
Abstract
Rail networks are key national infrastructure assets, providing mass transport capability for both goods and passengers. The reliability of these networks is therefore critical to avoid widescale disruption. To ensure reliability, they must be effectively monitored and maintained. The monitoring of the structural health of the Rail network is challenging with rolling stock of different weights and speeds travelling on the same lines. Current structural health monitoring techniques are unable to effectively monitor in real time defects forming on the rail. Therefore, to ensure continued reliability of the rail network new monitoring methods must be found.
This work investigates the use and advancement of acoustic emission techniques in monitoring the real time structural health of critical rail infrastructure. This work will focus on the monitoring of R220 and R260 grade steels used for plain track and cast manganese steel used for rail crossings.
Acoustic emission is already used across a wide number of industries for the detection of crack growth. This work initially looks at the feasibility of a commercially available system, procured from Physical Acoustics for the monitoring of fatigue crack growth in rail steels. Three key acoustic parameters were focussed on, Energy, Duration and Counts. A good correlation was found between increasing crack growth and these parameters with increasing crack severity for both the R220 and R260 steels, and to a lesser extent with the cast manganese steel. This correlation shows the potential for applying these techniques to the monitoring of fatigue crack growth in the rail environment.
It is proposed that these commercial systems are limited in their accuracy and capacity for real-time monitoring as the acoustic data is packaged into hits removing much of the available data. An alternative approach is therefore proposed using a customised acoustic emission monitoring system that captures and analysis the complete acoustic waveform.
The volume of data generated using the custom system necessitated the use of automated analysis techniques. Machine learning techniques were therefore developed in this work to analyse and classify the acoustic emission data generated during fatigue testing under laboratory conditions. Three signal processing techniques where tested; FFT, RMS and CWT with both shallow and deep neural networks developed for the FFT and RMS processing routes. High prediction accuracy was achieved using the custom system with the FFT shallow neural network achieving an accuracy of 87.8%.
Type of Work: | Thesis (Doctorates > Ph.D.) | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
Award Type: | Doctorates > Ph.D. | |||||||||
Supervisor(s): |
|
|||||||||
Licence: | All rights reserved | |||||||||
College/Faculty: | Colleges (2008 onwards) > College of Engineering & Physical Sciences | |||||||||
School or Department: | School of Metallurgy and Materials | |||||||||
Funders: | Engineering and Physical Sciences Research Council | |||||||||
Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science T Technology > TA Engineering (General). Civil engineering (General) T Technology > TF Railroad engineering and operation T Technology > TN Mining engineering. Metallurgy |
|||||||||
URI: | http://etheses.bham.ac.uk/id/eprint/13714 |
Actions
Request a Correction | |
View Item |
Downloads
Downloads per month over past year