eTheses Repository

Operational industrial fault detection and diagnosis: railway actuator case studies

Silmon, Joseph A. (2009)
Ph.D. thesis, University of Birmingham.

Loading
PDF (4Mb)

Abstract

Modern railways are required to operate with a high level of safety and reliability. The weakest components are those which have the highest safety requirements and the lowest inherent reliability. Single-throw mechanical actuators, such as powered train doors, trainstops, level crossing barriers and switch actuators (point machines) are a group of components which have these properties. Preventative maintenance is carried out periodically in order to mitigate the risks of these actuators failing. This is inefficient: a condition-based maintenance approach would reduce costs and the risks to staff. However, this kind of maintenance requires very accurate automatic condition monitoring. Currently, the threshold-based condition monitoring systems installed in pilot schemes around the country do not have enough insight into actuator performance to detect incipient faults. These are hard to spot because their symptoms develop over a long period of time. It is uneconomical to carry out detailed analysis or modelling, or collect a large amount of training data, for each instance of a large group of assets. Therefore, the solution needed to establish diagnosis rules based on offline analysis, or training data from only one actuator. This thesis draws on previous work in qualitative trend analysis to build a diagnosis system which uses a combined approach of qualitative and quantitative analysis to transfer the knowledge gathered from one actuator to its fellows in service. The method used has been designed to use straightforward components, so that it can be more easily explained to users. Two case studies were carried out in order to verify the system's functions. Data were collected from real-life actuators, under simulation of incipient faults. The diagnosis system then operated on the data. The system's performance was almost as good with real-world data as it was with synthetic data. The system has been a success when operating on the data gathered under laboratory conditions. In the real world, a system such as this could be used to post-process data gathered around the railway network from actuators with local data acquisition equipment. Incipient faults could be detected in the early stages of their development and accurately diagnosed, allowing maintenance effort to be targeted very specifically, saving money, time and exposing staff to fewer hazards.

Type of Work:Ph.D. thesis.
Supervisor(s):Roberts, Clive
School/Faculty:Colleges (2008 onwards) > College of Engineering & Physical Sciences
Department:Department of Electronic, Electrical and Computer Engineering
Keywords:Fault diagnosis, fuzzy set, railway, asset management, condition-based maintenance
Subjects:TK Electrical engineering. Electronics Nuclear engineering
TF Railroad engineering and operation
Institution:University of Birmingham
ID Code:481
This unpublished thesis/dissertation is copyright of the author and/or third parties. The intellectual property rights of the author or third parties in respect of this work are as defined by The Copyright Designs and Patents Act 1988 or as modified by any successor legislation. Any use made of information contained in this thesis/dissertation must be in accordance with that legislation and must be properly acknowledged. Further distribution or reproduction in any format is prohibited without the permission of the copyright holder.
Export Reference As : ASCII + BibTeX + Dublin Core + EndNote + HTML + METS + MODS + OpenURL Object + Reference Manager + Refer + RefWorks
Share this item :
QR Code for this page

Repository Staff Only: item control page