Reliability-as-a-Service for bearing risk assessment investigated with advanced mathematical models
Reliability-as-a-Service for bearing risk assessment investigated with advanced mathematical models
As a key player in bearing service life, the lubricant chemistry has a profound effect on bearing reliability. To increase the reliability of bearings, an Industrial Analytics solution is proposed for proactive condition monitoring and this is delivered via a Reliability-as-a-Service application. The performance predictions of bearings rely on customized algorithms with the main focus on digitalizing lubricant chemistry; the principles behind these processes are outlined in this study. Subsequently, independent testing is performed to confirm the ability of the presented Industrial Analytics solution for such predictions. By deciphering the chemical compounds of lubricants and characteristics of the interface, the Industrial Analytics solution delivers a precise bearing reliability assessment a priori to predict service life of the operation. Bearing tests have shown that the classification system of this Industrial Analytics solution is able to predict 12 out of 13 bearing failures (92%). The described approach provides a proactive bearing risk classification that allows the operator to take immediate action in reducing the failure potential during smooth operation - preventing any potential damage from occurring. For this purpose, a mathematical model is introduced that derives a set of classification rules for oil lubricants, based on linear binary classifiers (support vector machines) that are applied to the chemical compound's mixture data.
Bearing tribology, Classification, Industrial analytics, Reliability, Support vector machines
Brandt, Jan M.
4eface24-2bb1-4365-812c-82e8c096df00
Benedek, Márton
72bc97bc-a373-4b88-9e9f-145d62551214
Guerin, Jeffrey S.
754142e9-8ba7-4c97-8c52-23388794d083
Fliege, Jörg
54978787-a271-4f70-8494-3c701c893d98
14 March 2020
Brandt, Jan M.
4eface24-2bb1-4365-812c-82e8c096df00
Benedek, Márton
72bc97bc-a373-4b88-9e9f-145d62551214
Guerin, Jeffrey S.
754142e9-8ba7-4c97-8c52-23388794d083
Fliege, Jörg
54978787-a271-4f70-8494-3c701c893d98
Brandt, Jan M., Benedek, Márton, Guerin, Jeffrey S. and Fliege, Jörg
(2020)
Reliability-as-a-Service for bearing risk assessment investigated with advanced mathematical models.
Internet of Things (Netherlands), 11, [100178].
(doi:10.1016/j.iot.2020.100178).
Abstract
As a key player in bearing service life, the lubricant chemistry has a profound effect on bearing reliability. To increase the reliability of bearings, an Industrial Analytics solution is proposed for proactive condition monitoring and this is delivered via a Reliability-as-a-Service application. The performance predictions of bearings rely on customized algorithms with the main focus on digitalizing lubricant chemistry; the principles behind these processes are outlined in this study. Subsequently, independent testing is performed to confirm the ability of the presented Industrial Analytics solution for such predictions. By deciphering the chemical compounds of lubricants and characteristics of the interface, the Industrial Analytics solution delivers a precise bearing reliability assessment a priori to predict service life of the operation. Bearing tests have shown that the classification system of this Industrial Analytics solution is able to predict 12 out of 13 bearing failures (92%). The described approach provides a proactive bearing risk classification that allows the operator to take immediate action in reducing the failure potential during smooth operation - preventing any potential damage from occurring. For this purpose, a mathematical model is introduced that derives a set of classification rules for oil lubricants, based on linear binary classifiers (support vector machines) that are applied to the chemical compound's mixture data.
Text
1-s2.0-S2542660520300184-main
- Version of Record
More information
Accepted/In Press date: 10 February 2020
e-pub ahead of print date: 18 February 2020
Published date: 14 March 2020
Additional Information:
Funding Information:
This research was supported by the Bundesministerium für Wirtschaft und Energie (BMWi, Berlin, Germany; Grant Number 0324082); the second author would like to acknowledge support from the National Research, Development and Innovation Fund (TUDFO/51757/2019-ITM, Thematic Excellence Program) as well as the University of Southampton, UK. We would further like to thank Baher Azzam from the Center of Wind Power Drives (CWD), RWTH Aachen, Germany, for discussions on data analysis; we are also grateful to Amanda Belanger (IBM, Ottawa, ON, Canada), Steven Yankowich (General Dynamics, Ottawa, ON), and the Telfer School of Management, University of Ottawa, for the business insights and future aspects of Industrial Analytics and as-a-Services business models.
Keywords:
Bearing tribology, Classification, Industrial analytics, Reliability, Support vector machines
Identifiers
Local EPrints ID: 483852
URI: http://eprints.soton.ac.uk/id/eprint/483852
ISSN: 2542-6605
PURE UUID: ad6cb8a5-83d4-4b06-9b1c-69927d513d2b
Catalogue record
Date deposited: 07 Nov 2023 08:09
Last modified: 18 Mar 2024 03:08
Export record
Altmetrics
Contributors
Author:
Jan M. Brandt
Author:
Márton Benedek
Author:
Jeffrey S. Guerin
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics
