Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models

Bas M. Kessels; Rob H.B. Fey; Mohammad H. Abbasi; Nathan van de Wouw

doi:10.1007/978-3-031-04122-8_1

Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models

Bas M. Kessels, Rob H.B. Fey, Mohammad H. Abbasi, Nathan van de Wouw

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

3 Citations (Scopus)

9 Downloads (Pure)

Abstract

In order to ensure that a digital twin accurately describes the dynamic behavior of its corresponding physical system, model updating is typically applied. This chapter introduces a (near) real-time method that uses inverse mapping models to update first-principles-based nonlinear dynamics models. The inverse mapping model infers a set of physically interpretable updating parameter values on the basis of a set of time-domain features extracted from measurements on the real system. Here, the inverse model is given by an artificial neural network that is trained using simulated data. By using a simple nonlinear multibody model, it is illustrated that this method is able to accurately and precisely update parameter values with low computational effort.

Original language	English
Title of host publication	Data Science in Engineering
Subtitle of host publication	Proceedings of the 40th IMAC, A Conference and Exposition on Structural Dynamics, 2022
Editors	Ramin Madarshahian, Francois Hemez
Place of Publication	Cham
Publisher	Springer
Chapter	1
Pages	1-4
Number of pages	4
Volume	9
ISBN (Electronic)	978-3-031-04122-8
ISBN (Print)	978-3-031-04121-1
DOIs	https://doi.org/10.1007/978-3-031-04122-8_1
Publication status	Published - 2022
Event	40th IMAC, A Conference and Exposition on Structural Dynamics, 2022 - Rosen Plaza Hotel, Orlando, United States Duration: 7 Feb 2022 → 10 Feb 2022 Conference number: 40 https://sem.org/imac

Conference

Conference	40th IMAC, A Conference and Exposition on Structural Dynamics, 2022
Abbreviated title	IMAC-XL
Country/Territory	United States
City	Orlando
Period	7/02/22 → 10/02/22
Internet address	https://sem.org/imac

Funding

This publication is part of the project Digital Twin (project 2.1) with project number P18-03 of the research program Perspectief, which is (mainly) financed by the Dutch Research Council (NWO).

Keywords

Digital twin
Machine learning
Model updating
Nonlinear dynamics

Access to Document

10.1007/978-3-031-04122-8_1

pdfFinal published version, 364 KBLicence: TAVERNE

Cite this

Kessels, B. M., Fey, R. H. B., Abbasi, M. H., & van de Wouw, N. (2022). Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models. In R. Madarshahian, & F. Hemez (Eds.), Data Science in Engineering: Proceedings of the 40th IMAC, A Conference and Exposition on Structural Dynamics, 2022 (Vol. 9, pp. 1-4). Springer. https://doi.org/10.1007/978-3-031-04122-8_1

@inproceedings{9581504dc7c14ffa8efdda2c628b9740,

title = "Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models",

abstract = "In order to ensure that a digital twin accurately describes the dynamic behavior of its corresponding physical system, model updating is typically applied. This chapter introduces a (near) real-time method that uses inverse mapping models to update first-principles-based nonlinear dynamics models. The inverse mapping model infers a set of physically interpretable updating parameter values on the basis of a set of time-domain features extracted from measurements on the real system. Here, the inverse model is given by an artificial neural network that is trained using simulated data. By using a simple nonlinear multibody model, it is illustrated that this method is able to accurately and precisely update parameter values with low computational effort.",

keywords = "Digital twin, Machine learning, Model updating, Nonlinear dynamics",

author = "Kessels, \{Bas M.\} and Fey, \{Rob H.B.\} and Abbasi, \{Mohammad H.\} and \{van de Wouw\}, Nathan",

year = "2022",

doi = "10.1007/978-3-031-04122-8\_1",

language = "English",

isbn = "978-3-031-04121-1",

volume = "9",

pages = "1--4",

editor = "Ramin Madarshahian and Francois Hemez",

booktitle = "Data Science in Engineering",

publisher = "Springer",

address = "Germany",

note = "40th IMAC, A Conference and Exposition on Structural Dynamics, 2022, IMAC-XL ; Conference date: 07-02-2022 Through 10-02-2022",

url = "https://sem.org/imac",

}

Kessels, BM , Fey, RHB, Abbasi, MH & van de Wouw, N 2022, Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models. in R Madarshahian & F Hemez (eds), Data Science in Engineering: Proceedings of the 40th IMAC, A Conference and Exposition on Structural Dynamics, 2022. vol. 9, Springer, Cham, pp. 1-4, 40th IMAC, A Conference and Exposition on Structural Dynamics, 2022, Orlando, Florida, United States, 7/02/22. https://doi.org/10.1007/978-3-031-04122-8_1

Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models. / Kessels, Bas M.; Fey, Rob H.B.; Abbasi, Mohammad H. et al.
Data Science in Engineering: Proceedings of the 40th IMAC, A Conference and Exposition on Structural Dynamics, 2022. ed. / Ramin Madarshahian; Francois Hemez. Vol. 9 Cham: Springer, 2022. p. 1-4.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models

AU - Kessels, Bas M.

AU - Fey, Rob H.B.

AU - Abbasi, Mohammad H.

AU - van de Wouw, Nathan

N1 - Conference code: 40

PY - 2022

Y1 - 2022

N2 - In order to ensure that a digital twin accurately describes the dynamic behavior of its corresponding physical system, model updating is typically applied. This chapter introduces a (near) real-time method that uses inverse mapping models to update first-principles-based nonlinear dynamics models. The inverse mapping model infers a set of physically interpretable updating parameter values on the basis of a set of time-domain features extracted from measurements on the real system. Here, the inverse model is given by an artificial neural network that is trained using simulated data. By using a simple nonlinear multibody model, it is illustrated that this method is able to accurately and precisely update parameter values with low computational effort.

AB - In order to ensure that a digital twin accurately describes the dynamic behavior of its corresponding physical system, model updating is typically applied. This chapter introduces a (near) real-time method that uses inverse mapping models to update first-principles-based nonlinear dynamics models. The inverse mapping model infers a set of physically interpretable updating parameter values on the basis of a set of time-domain features extracted from measurements on the real system. Here, the inverse model is given by an artificial neural network that is trained using simulated data. By using a simple nonlinear multibody model, it is illustrated that this method is able to accurately and precisely update parameter values with low computational effort.

KW - Digital twin

KW - Machine learning

KW - Model updating

KW - Nonlinear dynamics

UR - https://www.scopus.com/pages/publications/85135083003

U2 - 10.1007/978-3-031-04122-8_1

DO - 10.1007/978-3-031-04122-8_1

M3 - Conference contribution

AN - SCOPUS:85135083003

SN - 978-3-031-04121-1

VL - 9

SP - 1

EP - 4

BT - Data Science in Engineering

A2 - Madarshahian, Ramin

A2 - Hemez, Francois

PB - Springer

CY - Cham

T2 - 40th IMAC, A Conference and Exposition on Structural Dynamics, 2022

Y2 - 7 February 2022 through 10 February 2022

ER -

Model Updating for Nonlinear Dynamic Digital Twins Using Data-Based Inverse Mapping Models

Abstract

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this