No Cover Image

Journal article 460 views 507 downloads

Machine learning based digital twin for dynamical systems with multiple time-scales

S. Chakraborty, Sondipon Adhikari

Computers & Structures, Volume: 243, Start page: 106410

Swansea University Author: Sondipon Adhikari

  • 55549.pdf

    PDF | Accepted Manuscript

    © 2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license

    Download (5.84MB)

Check full text

DOI (Published version): 10.1016/j.compstruc.2020.106410

Abstract

Digital twin technology has a huge potential for widespread applications in different industrial sectors such as infrastructure, aerospace, and automotive. However, practical adoptions of this technology have been slower, mainly due to a lack of application-specific details. Here we focus on a digit...

Full description

Published in: Computers & Structures
ISSN: 0045-7949
Published: Elsevier BV 2021
Online Access: Check full text

URI: https://cronfa.swan.ac.uk/Record/cronfa55549
Tags: Add Tag
No Tags, Be the first to tag this record!
first_indexed 2020-10-29T11:21:06Z
last_indexed 2020-12-05T04:09:41Z
id cronfa55549
recordtype SURis
fullrecord <?xml version="1.0"?><rfc1807><datestamp>2020-12-04T12:03:41.0062991</datestamp><bib-version>v2</bib-version><id>55549</id><entry>2020-10-29</entry><title>Machine learning based digital twin for dynamical systems with multiple time-scales</title><swanseaauthors><author><sid>4ea84d67c4e414f5ccbd7593a40f04d3</sid><firstname>Sondipon</firstname><surname>Adhikari</surname><name>Sondipon Adhikari</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2020-10-29</date><deptcode>FGSEN</deptcode><abstract>Digital twin technology has a huge potential for widespread applications in different industrial sectors such as infrastructure, aerospace, and automotive. However, practical adoptions of this technology have been slower, mainly due to a lack of application-specific details. Here we focus on a digital twin framework for linear single-degree-of-freedom structural dynamic systems evolving in two different operational time scales in addition to its intrinsic dynamic time-scale. Our approach strategically separates into two components &#x2013; (a) a physics-based nominal model for data processing and response predictions, and (b) a data-driven machine learning model for the time-evolution of the system parameters. The physics-based nominal model is system-specific and selected based on the problem under consideration. On the other hand, the data-driven machine learning model is generic. For tracking the multi-timescale evolution of the system parameters, we propose to exploit a mixture of experts as the data-driven model. Within the mixture of experts model, Gaussian Process (GP) is used as the expert model. The primary idea is to let each expert track the evolution of the system parameters at a single time-scale. For learning the hyperparameters of the &#x2018;mixture of experts using GP&#x2019;, an efficient framework that exploits expectation-maximization and sequential Monte Carlo sampler is used. Performance of the digital twin is illustrated on a multi-timescale dynamical system with stiffness and/or mass variations. The digital twin is found to be robust and yields reasonably accurate results. One exciting feature of the proposed digital twin is its capability to provide reasonable predictions at future time-steps. Aspects related to the data quality and data quantity are also investigated.</abstract><type>Journal Article</type><journal>Computers &amp; Structures</journal><volume>243</volume><journalNumber/><paginationStart>106410</paginationStart><paginationEnd/><publisher>Elsevier BV</publisher><placeOfPublication/><isbnPrint/><isbnElectronic/><issnPrint>0045-7949</issnPrint><issnElectronic/><keywords>Digital twin, Multi-timescale dynamics, Mixture of experts, Gaussian process, Frequency</keywords><publishedDay>15</publishedDay><publishedMonth>1</publishedMonth><publishedYear>2021</publishedYear><publishedDate>2021-01-15</publishedDate><doi>10.1016/j.compstruc.2020.106410</doi><url/><notes/><college>COLLEGE NANME</college><department>Science and Engineering - Faculty</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>FGSEN</DepartmentCode><institution>Swansea University</institution><apcterm/><lastEdited>2020-12-04T12:03:41.0062991</lastEdited><Created>2020-10-29T11:18:22.3984928</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Engineering and Applied Sciences - Uncategorised</level></path><authors><author><firstname>S.</firstname><surname>Chakraborty</surname><order>1</order></author><author><firstname>Sondipon</firstname><surname>Adhikari</surname><order>2</order></author></authors><documents><document><filename>55549__18535__88f4988a52984a849f651b45d48fea39.pdf</filename><originalFilename>55549.pdf</originalFilename><uploaded>2020-10-29T14:48:28.4729094</uploaded><type>Output</type><contentLength>6126292</contentLength><contentType>application/pdf</contentType><version>Accepted Manuscript</version><cronfaStatus>true</cronfaStatus><embargoDate>2021-10-23T00:00:00.0000000</embargoDate><documentNotes>&#xA9; 2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license</documentNotes><copyrightCorrect>true</copyrightCorrect><language>eng</language><licence>http://creativecommons.org/licenses/by-nc-nd/4.0/</licence></document></documents><OutputDurs/></rfc1807>
spelling 2020-12-04T12:03:41.0062991 v2 55549 2020-10-29 Machine learning based digital twin for dynamical systems with multiple time-scales 4ea84d67c4e414f5ccbd7593a40f04d3 Sondipon Adhikari Sondipon Adhikari true false 2020-10-29 FGSEN Digital twin technology has a huge potential for widespread applications in different industrial sectors such as infrastructure, aerospace, and automotive. However, practical adoptions of this technology have been slower, mainly due to a lack of application-specific details. Here we focus on a digital twin framework for linear single-degree-of-freedom structural dynamic systems evolving in two different operational time scales in addition to its intrinsic dynamic time-scale. Our approach strategically separates into two components – (a) a physics-based nominal model for data processing and response predictions, and (b) a data-driven machine learning model for the time-evolution of the system parameters. The physics-based nominal model is system-specific and selected based on the problem under consideration. On the other hand, the data-driven machine learning model is generic. For tracking the multi-timescale evolution of the system parameters, we propose to exploit a mixture of experts as the data-driven model. Within the mixture of experts model, Gaussian Process (GP) is used as the expert model. The primary idea is to let each expert track the evolution of the system parameters at a single time-scale. For learning the hyperparameters of the ‘mixture of experts using GP’, an efficient framework that exploits expectation-maximization and sequential Monte Carlo sampler is used. Performance of the digital twin is illustrated on a multi-timescale dynamical system with stiffness and/or mass variations. The digital twin is found to be robust and yields reasonably accurate results. One exciting feature of the proposed digital twin is its capability to provide reasonable predictions at future time-steps. Aspects related to the data quality and data quantity are also investigated. Journal Article Computers & Structures 243 106410 Elsevier BV 0045-7949 Digital twin, Multi-timescale dynamics, Mixture of experts, Gaussian process, Frequency 15 1 2021 2021-01-15 10.1016/j.compstruc.2020.106410 COLLEGE NANME Science and Engineering - Faculty COLLEGE CODE FGSEN Swansea University 2020-12-04T12:03:41.0062991 2020-10-29T11:18:22.3984928 Faculty of Science and Engineering School of Engineering and Applied Sciences - Uncategorised S. Chakraborty 1 Sondipon Adhikari 2 55549__18535__88f4988a52984a849f651b45d48fea39.pdf 55549.pdf 2020-10-29T14:48:28.4729094 Output 6126292 application/pdf Accepted Manuscript true 2021-10-23T00:00:00.0000000 © 2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license true eng http://creativecommons.org/licenses/by-nc-nd/4.0/
title Machine learning based digital twin for dynamical systems with multiple time-scales
spellingShingle Machine learning based digital twin for dynamical systems with multiple time-scales
Sondipon Adhikari
title_short Machine learning based digital twin for dynamical systems with multiple time-scales
title_full Machine learning based digital twin for dynamical systems with multiple time-scales
title_fullStr Machine learning based digital twin for dynamical systems with multiple time-scales
title_full_unstemmed Machine learning based digital twin for dynamical systems with multiple time-scales
title_sort Machine learning based digital twin for dynamical systems with multiple time-scales
author_id_str_mv 4ea84d67c4e414f5ccbd7593a40f04d3
author_id_fullname_str_mv 4ea84d67c4e414f5ccbd7593a40f04d3_***_Sondipon Adhikari
author Sondipon Adhikari
author2 S. Chakraborty
Sondipon Adhikari
format Journal article
container_title Computers & Structures
container_volume 243
container_start_page 106410
publishDate 2021
institution Swansea University
issn 0045-7949
doi_str_mv 10.1016/j.compstruc.2020.106410
publisher Elsevier BV
college_str Faculty of Science and Engineering
hierarchytype
hierarchy_top_id facultyofscienceandengineering
hierarchy_top_title Faculty of Science and Engineering
hierarchy_parent_id facultyofscienceandengineering
hierarchy_parent_title Faculty of Science and Engineering
department_str School of Engineering and Applied Sciences - Uncategorised{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Engineering and Applied Sciences - Uncategorised
document_store_str 1
active_str 0
description Digital twin technology has a huge potential for widespread applications in different industrial sectors such as infrastructure, aerospace, and automotive. However, practical adoptions of this technology have been slower, mainly due to a lack of application-specific details. Here we focus on a digital twin framework for linear single-degree-of-freedom structural dynamic systems evolving in two different operational time scales in addition to its intrinsic dynamic time-scale. Our approach strategically separates into two components – (a) a physics-based nominal model for data processing and response predictions, and (b) a data-driven machine learning model for the time-evolution of the system parameters. The physics-based nominal model is system-specific and selected based on the problem under consideration. On the other hand, the data-driven machine learning model is generic. For tracking the multi-timescale evolution of the system parameters, we propose to exploit a mixture of experts as the data-driven model. Within the mixture of experts model, Gaussian Process (GP) is used as the expert model. The primary idea is to let each expert track the evolution of the system parameters at a single time-scale. For learning the hyperparameters of the ‘mixture of experts using GP’, an efficient framework that exploits expectation-maximization and sequential Monte Carlo sampler is used. Performance of the digital twin is illustrated on a multi-timescale dynamical system with stiffness and/or mass variations. The digital twin is found to be robust and yields reasonably accurate results. One exciting feature of the proposed digital twin is its capability to provide reasonable predictions at future time-steps. Aspects related to the data quality and data quantity are also investigated.
published_date 2021-01-15T04:09:50Z
_version_ 1763753686142550016
score 11.013194

Similar Items