Gaussian process time series model for life prognosis of metallic structures

Subhasish Mohanty, Santanu Das, Aditi Chattopadhyay, Pedro Peralta

Research output: Contribution to journalArticle

34 Citations (Scopus)

Abstract

Al 2024-T351 fatigue specimens have been modeled using a kernel-based multi-variate Gaussian Process approach. The Gaussian Process model projects fatigue affecting input variables to output crack growth by probabilistically inferring the underlying nonlinear relationship between input and output. The Gaussian Process approach not only explicitly models the uncertainty due to scatter in material microstructure parameter but it also implicitly models the loading sequence effect due to variable loading. The loading sequence effect is modeled through the Gaussian Process optimal hyperparameters by using the crack length data observed over the entire domain of spectrum loading. The performance in the crack growth prediction is evaluated for two covariance functions, a radial basis-based, anisotropic, covariance function and a neural network-based isotropic covariance function. Furthermore, the performance of different types of scaling, used to scale the inputĝ€"output data space, is tested. It is found that for the radial basis-based anisotropic covariance function with normalized scaling, the prediction error is consistently lower compared to other combinations. In addition, the Gaussian Process model allows determination of the collapse load condition, which is a desirable feature for the online health monitoring and prognosis.

Original languageEnglish (US)
Pages (from-to)887-896
Number of pages10
JournalJournal of Intelligent Material Systems and Structures
Volume20
Issue number8
DOIs
StatePublished - May 2009

Fingerprint

Time series
Crack propagation
Fatigue of materials
Health
Cracks
Neural networks
Microstructure
Monitoring

Keywords

  • 2024-T351 aluminum alloy
  • Covariance function
  • Fatigue crack growth
  • Gaussian Process
  • Hyperparameters.
  • Maximum likelihood optimization
  • Prognosis
  • Variable loading

ASJC Scopus subject areas

  • Materials Science(all)
  • Mechanical Engineering

Cite this

Gaussian process time series model for life prognosis of metallic structures. / Mohanty, Subhasish; Das, Santanu; Chattopadhyay, Aditi; Peralta, Pedro.

In: Journal of Intelligent Material Systems and Structures, Vol. 20, No. 8, 05.2009, p. 887-896.

Research output: Contribution to journalArticle

@article{eaf54ef66d574c4ab02c8981989f9f0b,
title = "Gaussian process time series model for life prognosis of metallic structures",
abstract = "Al 2024-T351 fatigue specimens have been modeled using a kernel-based multi-variate Gaussian Process approach. The Gaussian Process model projects fatigue affecting input variables to output crack growth by probabilistically inferring the underlying nonlinear relationship between input and output. The Gaussian Process approach not only explicitly models the uncertainty due to scatter in material microstructure parameter but it also implicitly models the loading sequence effect due to variable loading. The loading sequence effect is modeled through the Gaussian Process optimal hyperparameters by using the crack length data observed over the entire domain of spectrum loading. The performance in the crack growth prediction is evaluated for two covariance functions, a radial basis-based, anisotropic, covariance function and a neural network-based isotropic covariance function. Furthermore, the performance of different types of scaling, used to scale the inputĝ€{"}output data space, is tested. It is found that for the radial basis-based anisotropic covariance function with normalized scaling, the prediction error is consistently lower compared to other combinations. In addition, the Gaussian Process model allows determination of the collapse load condition, which is a desirable feature for the online health monitoring and prognosis.",
keywords = "2024-T351 aluminum alloy, Covariance function, Fatigue crack growth, Gaussian Process, Hyperparameters., Maximum likelihood optimization, Prognosis, Variable loading",
author = "Subhasish Mohanty and Santanu Das and Aditi Chattopadhyay and Pedro Peralta",
year = "2009",
month = "5",
doi = "10.1177/1045389X08099602",
language = "English (US)",
volume = "20",
pages = "887--896",
journal = "Journal of Intelligent Material Systems and Structures",
issn = "1045-389X",
publisher = "SAGE Publications Ltd",
number = "8",

}

TY - JOUR

T1 - Gaussian process time series model for life prognosis of metallic structures

AU - Mohanty, Subhasish

AU - Das, Santanu

AU - Chattopadhyay, Aditi

AU - Peralta, Pedro

PY - 2009/5

Y1 - 2009/5

N2 - Al 2024-T351 fatigue specimens have been modeled using a kernel-based multi-variate Gaussian Process approach. The Gaussian Process model projects fatigue affecting input variables to output crack growth by probabilistically inferring the underlying nonlinear relationship between input and output. The Gaussian Process approach not only explicitly models the uncertainty due to scatter in material microstructure parameter but it also implicitly models the loading sequence effect due to variable loading. The loading sequence effect is modeled through the Gaussian Process optimal hyperparameters by using the crack length data observed over the entire domain of spectrum loading. The performance in the crack growth prediction is evaluated for two covariance functions, a radial basis-based, anisotropic, covariance function and a neural network-based isotropic covariance function. Furthermore, the performance of different types of scaling, used to scale the inputĝ€"output data space, is tested. It is found that for the radial basis-based anisotropic covariance function with normalized scaling, the prediction error is consistently lower compared to other combinations. In addition, the Gaussian Process model allows determination of the collapse load condition, which is a desirable feature for the online health monitoring and prognosis.

AB - Al 2024-T351 fatigue specimens have been modeled using a kernel-based multi-variate Gaussian Process approach. The Gaussian Process model projects fatigue affecting input variables to output crack growth by probabilistically inferring the underlying nonlinear relationship between input and output. The Gaussian Process approach not only explicitly models the uncertainty due to scatter in material microstructure parameter but it also implicitly models the loading sequence effect due to variable loading. The loading sequence effect is modeled through the Gaussian Process optimal hyperparameters by using the crack length data observed over the entire domain of spectrum loading. The performance in the crack growth prediction is evaluated for two covariance functions, a radial basis-based, anisotropic, covariance function and a neural network-based isotropic covariance function. Furthermore, the performance of different types of scaling, used to scale the inputĝ€"output data space, is tested. It is found that for the radial basis-based anisotropic covariance function with normalized scaling, the prediction error is consistently lower compared to other combinations. In addition, the Gaussian Process model allows determination of the collapse load condition, which is a desirable feature for the online health monitoring and prognosis.

KW - 2024-T351 aluminum alloy

KW - Covariance function

KW - Fatigue crack growth

KW - Gaussian Process

KW - Hyperparameters.

KW - Maximum likelihood optimization

KW - Prognosis

KW - Variable loading

UR - http://www.scopus.com/inward/record.url?scp=65649153376&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=65649153376&partnerID=8YFLogxK

U2 - 10.1177/1045389X08099602

DO - 10.1177/1045389X08099602

M3 - Article

AN - SCOPUS:65649153376

VL - 20

SP - 887

EP - 896

JO - Journal of Intelligent Material Systems and Structures

JF - Journal of Intelligent Material Systems and Structures

SN - 1045-389X

IS - 8

ER -