On Analysis of the Predictive Maintenance of Railway Points Processes and Possibilities
Loading...
URL
Journal Title
Journal ISSN
Volume Title
Perustieteiden korkeakoulu |
Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Authors
Date
2018-05-14
Department
Major/Subject
Machine Learning and Data Mining
Mcode
SCI3044
Degree programme
Master’s Programme in Computer, Communication and Information Sciences
Language
en
Pages
73+11
Series
Abstract
A railway network is vital condition for a blooming industry and fluent public transportation in most countries. To maintain safety and fluency in the traffic it needs to be constantly repaired. A pivotal part of the network -- railway points and their maintenance actions, is heavily regulated, leading to periodical visits to the points. However, these visit do not prevent all failures in the railway points and additionally are very costly. Scientists are constantly seeking possibilities to achieve condition-based regulation. However, all the reasonable approaches studied require both a lot of investments in additional equipment and the co-operation of several companies (those responsible for different aspects of the network). Recently cloud-based digitalisation in the railway industry has made multi company co-operation more practical and brought possibilities for data analysis. This thesis describes three aspects that are necessary to gain more from digitalisation in this field. First, principles and challenges of data analysis project. Secondly thesis investigates feasibility of railway point failures prediction between periodical maintenance visits with existing data. Thirdly, company co-operation requirements regarding data quality, and the formats of signalling logs and maintenance reports. The emphasis is on feasibility analysis and, with the help of typical machine learning algorithms, this thesis shows that there is potential to improve maintenance planning with existing data. However, the prediction accuracies achieved in the thesis indicates that without investing in additional equipment or more precise log measures, the accuracies are not in correct level to start processes towards condition-based regulation.Description
Supervisor
Vehtari, AkiThesis advisor
Eskelinen, JyriKeywords
predictive maintenance, data mining process, feasibility analysis, railway point maintenance, log analysis, machine learning