Optimal selection of autoregressive model coefficients for early damage detectability with an application to wind turbine blades

Simon Hoell, Piotr Omenzetter

Research output: Contribution to journalArticle

20 Citations (Scopus)
5 Downloads (Pure)

Abstract

Data-driven vibration-based damage detection techniques can be competitive because of their lower instrumentation and data analysis costs. The use of autoregressive model coefficients (ARMCs) as damage sensitive features (DSFs) is one such technique. So far, like with other DSFs, either full sets of coefficients or subsets selected by trial-and-error have been used, but this can lead to suboptimal composition of multivariate DSFs and decreased damage detection performance. This study enhances the selection of ARMCs for statistical hypothesis testing for damage presence. Two approaches for systematic ARMC selection, based on either adding or eliminating the coefficients one by one or using a genetic algorithm (GA) are proposed. The methods are applied to a numerical model of an aerodynamically excited large composite wind turbine blade with disbonding damage. The GA out performs the other selection methods and enables building multivariate DSFs that markedly enhance early damage detectability and are insensitive to measurement noise.
Original languageEnglish
Pages (from-to)557-577
Number of pages21
JournalMechanical Systems and Signal Processing
Volume70-71
Early online date26 Sep 2015
DOIs
Publication statusPublished - Mar 2016

Fingerprint

Wind turbines
Turbomachine blades
Damage detection
Genetic algorithms
Numerical models
Composite materials
Testing
Chemical analysis
Costs

Keywords

  • autoregressive models
  • damage detection
  • hypothesis testing
  • optimal feature selection
  • time series
  • wind turbine

Cite this

@article{79aabf6d6e394d54af720bd7b1efb2df,
title = "Optimal selection of autoregressive model coefficients for early damage detectability with an application to wind turbine blades",
abstract = "Data-driven vibration-based damage detection techniques can be competitive because of their lower instrumentation and data analysis costs. The use of autoregressive model coefficients (ARMCs) as damage sensitive features (DSFs) is one such technique. So far, like with other DSFs, either full sets of coefficients or subsets selected by trial-and-error have been used, but this can lead to suboptimal composition of multivariate DSFs and decreased damage detection performance. This study enhances the selection of ARMCs for statistical hypothesis testing for damage presence. Two approaches for systematic ARMC selection, based on either adding or eliminating the coefficients one by one or using a genetic algorithm (GA) are proposed. The methods are applied to a numerical model of an aerodynamically excited large composite wind turbine blade with disbonding damage. The GA out performs the other selection methods and enables building multivariate DSFs that markedly enhance early damage detectability and are insensitive to measurement noise.",
keywords = "autoregressive models , damage detection, hypothesis testing, optimal feature selection, time series, wind turbine",
author = "Simon Hoell and Piotr Omenzetter",
note = "Piotr Omenzetter and Simon Hoell's work within the Lloyd's Register Foundation Centre for Safety and Reliability Engineering at the University of Aberdeen is supported by Lloyd’s Register Foundation. The Foundation helps to protect life and property by supporting engineering-related education, public engagement and the application of research.",
year = "2016",
month = "3",
doi = "10.1016/j.ymssp.2015.09.007",
language = "English",
volume = "70-71",
pages = "557--577",
journal = "Mechanical Systems and Signal Processing",
issn = "0888-3270",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Optimal selection of autoregressive model coefficients for early damage detectability with an application to wind turbine blades

AU - Hoell, Simon

AU - Omenzetter, Piotr

N1 - Piotr Omenzetter and Simon Hoell's work within the Lloyd's Register Foundation Centre for Safety and Reliability Engineering at the University of Aberdeen is supported by Lloyd’s Register Foundation. The Foundation helps to protect life and property by supporting engineering-related education, public engagement and the application of research.

PY - 2016/3

Y1 - 2016/3

N2 - Data-driven vibration-based damage detection techniques can be competitive because of their lower instrumentation and data analysis costs. The use of autoregressive model coefficients (ARMCs) as damage sensitive features (DSFs) is one such technique. So far, like with other DSFs, either full sets of coefficients or subsets selected by trial-and-error have been used, but this can lead to suboptimal composition of multivariate DSFs and decreased damage detection performance. This study enhances the selection of ARMCs for statistical hypothesis testing for damage presence. Two approaches for systematic ARMC selection, based on either adding or eliminating the coefficients one by one or using a genetic algorithm (GA) are proposed. The methods are applied to a numerical model of an aerodynamically excited large composite wind turbine blade with disbonding damage. The GA out performs the other selection methods and enables building multivariate DSFs that markedly enhance early damage detectability and are insensitive to measurement noise.

AB - Data-driven vibration-based damage detection techniques can be competitive because of their lower instrumentation and data analysis costs. The use of autoregressive model coefficients (ARMCs) as damage sensitive features (DSFs) is one such technique. So far, like with other DSFs, either full sets of coefficients or subsets selected by trial-and-error have been used, but this can lead to suboptimal composition of multivariate DSFs and decreased damage detection performance. This study enhances the selection of ARMCs for statistical hypothesis testing for damage presence. Two approaches for systematic ARMC selection, based on either adding or eliminating the coefficients one by one or using a genetic algorithm (GA) are proposed. The methods are applied to a numerical model of an aerodynamically excited large composite wind turbine blade with disbonding damage. The GA out performs the other selection methods and enables building multivariate DSFs that markedly enhance early damage detectability and are insensitive to measurement noise.

KW - autoregressive models

KW - damage detection

KW - hypothesis testing

KW - optimal feature selection

KW - time series

KW - wind turbine

U2 - 10.1016/j.ymssp.2015.09.007

DO - 10.1016/j.ymssp.2015.09.007

M3 - Article

VL - 70-71

SP - 557

EP - 577

JO - Mechanical Systems and Signal Processing

JF - Mechanical Systems and Signal Processing

SN - 0888-3270

ER -