Efficient dimension reduction and surrogate-based sensitivity analysis for expensive models with high-dimensional outputs

Min Li, Ruo Qian Wang, Gaofeng Jia

Research output: Contribution to journalArticle

Abstract

Sensitivity analysis has been widely used to gain more insights on complex system behavior, to facilitate model reduction, system design and decision making. Typically, sensitivity analysis entails many evaluations of the system model. For expensive system models with high-dimensional outputs, direct adoption of such models for sensitivity analysis poses significant challenges in computational effort and memory requirements. To address these challenges, this paper proposes an efficient sensitivity analysis approach. The proposed method uses surrogate model to replace the expensive model for sensitivity analysis, and tackle the problem of building surrogate models for high-dimensional outputs through surrogate model integrated with dimension reduction. More specifically, the proposed method first uses surrogate models in low-dimensional latent output space to efficiently calculate the relevant covariance matrices for the low-dimensional latent outputs, and then directly establishes the sensitivity indices for the original high-dimensional output based on these covariance matrices and the derived transformation. Two examples are presented to demonstrate the efficiency and accuracy of the proposed method.

Original languageEnglish (US)
Article number106725
JournalReliability Engineering and System Safety
Volume195
DOIs
StatePublished - Mar 2020

Fingerprint

Sensitivity analysis
Covariance matrix
Large scale systems
Decision making
Systems analysis
Data storage equipment

All Science Journal Classification (ASJC) codes

  • Safety, Risk, Reliability and Quality
  • Industrial and Manufacturing Engineering

Keywords

  • Dimension reduction
  • High-dimensional outputs
  • Kriging surrogate model
  • Principal component analysis
  • Sensitivity analysis
  • Sobol’ index

Cite this

@article{644ba402ddd3479285071dce69faeb88,
title = "Efficient dimension reduction and surrogate-based sensitivity analysis for expensive models with high-dimensional outputs",
abstract = "Sensitivity analysis has been widely used to gain more insights on complex system behavior, to facilitate model reduction, system design and decision making. Typically, sensitivity analysis entails many evaluations of the system model. For expensive system models with high-dimensional outputs, direct adoption of such models for sensitivity analysis poses significant challenges in computational effort and memory requirements. To address these challenges, this paper proposes an efficient sensitivity analysis approach. The proposed method uses surrogate model to replace the expensive model for sensitivity analysis, and tackle the problem of building surrogate models for high-dimensional outputs through surrogate model integrated with dimension reduction. More specifically, the proposed method first uses surrogate models in low-dimensional latent output space to efficiently calculate the relevant covariance matrices for the low-dimensional latent outputs, and then directly establishes the sensitivity indices for the original high-dimensional output based on these covariance matrices and the derived transformation. Two examples are presented to demonstrate the efficiency and accuracy of the proposed method.",
keywords = "Dimension reduction, High-dimensional outputs, Kriging surrogate model, Principal component analysis, Sensitivity analysis, Sobol’ index",
author = "Min Li and Wang, {Ruo Qian} and Gaofeng Jia",
year = "2020",
month = "3",
doi = "10.1016/j.ress.2019.106725",
language = "English (US)",
volume = "195",
journal = "Reliability Engineering and System Safety",
issn = "0951-8320",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Efficient dimension reduction and surrogate-based sensitivity analysis for expensive models with high-dimensional outputs

AU - Li, Min

AU - Wang, Ruo Qian

AU - Jia, Gaofeng

PY - 2020/3

Y1 - 2020/3

N2 - Sensitivity analysis has been widely used to gain more insights on complex system behavior, to facilitate model reduction, system design and decision making. Typically, sensitivity analysis entails many evaluations of the system model. For expensive system models with high-dimensional outputs, direct adoption of such models for sensitivity analysis poses significant challenges in computational effort and memory requirements. To address these challenges, this paper proposes an efficient sensitivity analysis approach. The proposed method uses surrogate model to replace the expensive model for sensitivity analysis, and tackle the problem of building surrogate models for high-dimensional outputs through surrogate model integrated with dimension reduction. More specifically, the proposed method first uses surrogate models in low-dimensional latent output space to efficiently calculate the relevant covariance matrices for the low-dimensional latent outputs, and then directly establishes the sensitivity indices for the original high-dimensional output based on these covariance matrices and the derived transformation. Two examples are presented to demonstrate the efficiency and accuracy of the proposed method.

AB - Sensitivity analysis has been widely used to gain more insights on complex system behavior, to facilitate model reduction, system design and decision making. Typically, sensitivity analysis entails many evaluations of the system model. For expensive system models with high-dimensional outputs, direct adoption of such models for sensitivity analysis poses significant challenges in computational effort and memory requirements. To address these challenges, this paper proposes an efficient sensitivity analysis approach. The proposed method uses surrogate model to replace the expensive model for sensitivity analysis, and tackle the problem of building surrogate models for high-dimensional outputs through surrogate model integrated with dimension reduction. More specifically, the proposed method first uses surrogate models in low-dimensional latent output space to efficiently calculate the relevant covariance matrices for the low-dimensional latent outputs, and then directly establishes the sensitivity indices for the original high-dimensional output based on these covariance matrices and the derived transformation. Two examples are presented to demonstrate the efficiency and accuracy of the proposed method.

KW - Dimension reduction

KW - High-dimensional outputs

KW - Kriging surrogate model

KW - Principal component analysis

KW - Sensitivity analysis

KW - Sobol’ index

UR - http://www.scopus.com/inward/record.url?scp=85074668197&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074668197&partnerID=8YFLogxK

U2 - 10.1016/j.ress.2019.106725

DO - 10.1016/j.ress.2019.106725

M3 - Article

AN - SCOPUS:85074668197

VL - 195

JO - Reliability Engineering and System Safety

JF - Reliability Engineering and System Safety

SN - 0951-8320

M1 - 106725

ER -