Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations

Simon Hoell, Piotr Omenzetter

Research output: Chapter in Book/Report/Conference proceedingChapter (peer-reviewed)

Abstract

Potential energy outputs of wind turbines (WTs) are subject to continuous enhancements due to increasing demands for carbon neutral energy. The use of novel composite materials facilitates erections of ever larger WTs, which capture more energy by using longer WT blades (WTBs) with reduced weight. However, higher flexibilities and lower buckling capacities of these WTBs adversely affect long-term safety and reliability of WTs and, with it, energy production costs. This can be counteracted with the help of efficient structural health monitoring (SHM). The present study shows a novel methodology for vibration-based structural damage detection and severity estimation in WTBs. First, correlations of vibrational response signals are extracted as initial damage sensitive features (DSFs). Second, the Fukunaga-Koontz transform, an extension of the better-known Karhunen-Loéve expansion, is applied for extracting secondary DSFs with improved damage sensitivities. Third, univariate rankings of both DSFs are separately created with respect to the area under the receiver operating characteristic curve. Then, structural damage detection and severity estimation is performed with the help of hierarchical adaptive neuro-fuzzy inference systems, where the hierarchical structure allows accounting for the ranking information. The method is applied to laboratory experimental data from a small WTB exited by an air stream produced by a household fan. Damage severity estimation is studied by attaching different small masses as non-destructive damage scenarios. The results demonstrate that the proposed methodology enables to detect and estimate accurately the severity of the simulated damage. Furthermore, the advantages of using transformed DSFs are shown. This is promising for future developments of vibration-based SHM to facilitate improved safety and reliability of WTs at lower costs.

Original languageEnglish
Title of host publication8th European Workshop on Structural Health Monitoring, EWSHM 2016
PublisherNDT.net
Pages2281-2290
Number of pages10
Volume3
ISBN (Electronic)9781510827936
Publication statusPublished - 2016
Event8th European Workshop on Structural Health Monitoring, EWSHM 2016 - Bilbao, Spain
Duration: 5 Jul 20168 Jul 2016

Conference

Conference8th European Workshop on Structural Health Monitoring, EWSHM 2016
CountrySpain
CityBilbao
Period5/07/168/07/16

Fingerprint

Vibration
Wind turbines
Turbomachine blades
Damage detection
Structural health monitoring
Safety
Costs and Cost Analysis
Love
Health
Fuzzy inference
Potential energy
ROC Curve
Fans
Buckling
Costs
Carbon
Air
Weights and Measures
Composite materials

Keywords

  • Damage detection
  • Damage severity estimation
  • Fukunaga-Koontz transform
  • Fuzzy computing
  • Neural networks
  • Time series methods
  • Wind turbines

ASJC Scopus subject areas

  • Health Information Management
  • Computer Science Applications

Cite this

Hoell, S., & Omenzetter, P. (2016). Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations. In 8th European Workshop on Structural Health Monitoring, EWSHM 2016 (Vol. 3, pp. 2281-2290). NDT.net.

Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations. / Hoell, Simon; Omenzetter, Piotr.

8th European Workshop on Structural Health Monitoring, EWSHM 2016. Vol. 3 NDT.net, 2016. p. 2281-2290.

Research output: Chapter in Book/Report/Conference proceedingChapter (peer-reviewed)

Hoell, S & Omenzetter, P 2016, Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations. in 8th European Workshop on Structural Health Monitoring, EWSHM 2016. vol. 3, NDT.net, pp. 2281-2290, 8th European Workshop on Structural Health Monitoring, EWSHM 2016, Bilbao, Spain, 5/07/16.
Hoell S, Omenzetter P. Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations. In 8th European Workshop on Structural Health Monitoring, EWSHM 2016. Vol. 3. NDT.net. 2016. p. 2281-2290
Hoell, Simon ; Omenzetter, Piotr. / Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations. 8th European Workshop on Structural Health Monitoring, EWSHM 2016. Vol. 3 NDT.net, 2016. pp. 2281-2290
@inbook{e6b5ee2942a543f69f452bb21823da22,
title = "Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations",
abstract = "Potential energy outputs of wind turbines (WTs) are subject to continuous enhancements due to increasing demands for carbon neutral energy. The use of novel composite materials facilitates erections of ever larger WTs, which capture more energy by using longer WT blades (WTBs) with reduced weight. However, higher flexibilities and lower buckling capacities of these WTBs adversely affect long-term safety and reliability of WTs and, with it, energy production costs. This can be counteracted with the help of efficient structural health monitoring (SHM). The present study shows a novel methodology for vibration-based structural damage detection and severity estimation in WTBs. First, correlations of vibrational response signals are extracted as initial damage sensitive features (DSFs). Second, the Fukunaga-Koontz transform, an extension of the better-known Karhunen-Lo{\'e}ve expansion, is applied for extracting secondary DSFs with improved damage sensitivities. Third, univariate rankings of both DSFs are separately created with respect to the area under the receiver operating characteristic curve. Then, structural damage detection and severity estimation is performed with the help of hierarchical adaptive neuro-fuzzy inference systems, where the hierarchical structure allows accounting for the ranking information. The method is applied to laboratory experimental data from a small WTB exited by an air stream produced by a household fan. Damage severity estimation is studied by attaching different small masses as non-destructive damage scenarios. The results demonstrate that the proposed methodology enables to detect and estimate accurately the severity of the simulated damage. Furthermore, the advantages of using transformed DSFs are shown. This is promising for future developments of vibration-based SHM to facilitate improved safety and reliability of WTs at lower costs.",
keywords = "Damage detection, Damage severity estimation, Fukunaga-Koontz transform, Fuzzy computing, Neural networks, Time series methods, Wind turbines",
author = "Simon Hoell and Piotr Omenzetter",
year = "2016",
language = "English",
volume = "3",
pages = "2281--2290",
booktitle = "8th European Workshop on Structural Health Monitoring, EWSHM 2016",
publisher = "NDT.net",

}

TY - CHAP

T1 - Efficient neuro-fuzzy damage severity estimation in an experimental wind turbine blade using the Fukunaga-Koontz transform of vibration signal correlations

AU - Hoell, Simon

AU - Omenzetter, Piotr

PY - 2016

Y1 - 2016

N2 - Potential energy outputs of wind turbines (WTs) are subject to continuous enhancements due to increasing demands for carbon neutral energy. The use of novel composite materials facilitates erections of ever larger WTs, which capture more energy by using longer WT blades (WTBs) with reduced weight. However, higher flexibilities and lower buckling capacities of these WTBs adversely affect long-term safety and reliability of WTs and, with it, energy production costs. This can be counteracted with the help of efficient structural health monitoring (SHM). The present study shows a novel methodology for vibration-based structural damage detection and severity estimation in WTBs. First, correlations of vibrational response signals are extracted as initial damage sensitive features (DSFs). Second, the Fukunaga-Koontz transform, an extension of the better-known Karhunen-Loéve expansion, is applied for extracting secondary DSFs with improved damage sensitivities. Third, univariate rankings of both DSFs are separately created with respect to the area under the receiver operating characteristic curve. Then, structural damage detection and severity estimation is performed with the help of hierarchical adaptive neuro-fuzzy inference systems, where the hierarchical structure allows accounting for the ranking information. The method is applied to laboratory experimental data from a small WTB exited by an air stream produced by a household fan. Damage severity estimation is studied by attaching different small masses as non-destructive damage scenarios. The results demonstrate that the proposed methodology enables to detect and estimate accurately the severity of the simulated damage. Furthermore, the advantages of using transformed DSFs are shown. This is promising for future developments of vibration-based SHM to facilitate improved safety and reliability of WTs at lower costs.

AB - Potential energy outputs of wind turbines (WTs) are subject to continuous enhancements due to increasing demands for carbon neutral energy. The use of novel composite materials facilitates erections of ever larger WTs, which capture more energy by using longer WT blades (WTBs) with reduced weight. However, higher flexibilities and lower buckling capacities of these WTBs adversely affect long-term safety and reliability of WTs and, with it, energy production costs. This can be counteracted with the help of efficient structural health monitoring (SHM). The present study shows a novel methodology for vibration-based structural damage detection and severity estimation in WTBs. First, correlations of vibrational response signals are extracted as initial damage sensitive features (DSFs). Second, the Fukunaga-Koontz transform, an extension of the better-known Karhunen-Loéve expansion, is applied for extracting secondary DSFs with improved damage sensitivities. Third, univariate rankings of both DSFs are separately created with respect to the area under the receiver operating characteristic curve. Then, structural damage detection and severity estimation is performed with the help of hierarchical adaptive neuro-fuzzy inference systems, where the hierarchical structure allows accounting for the ranking information. The method is applied to laboratory experimental data from a small WTB exited by an air stream produced by a household fan. Damage severity estimation is studied by attaching different small masses as non-destructive damage scenarios. The results demonstrate that the proposed methodology enables to detect and estimate accurately the severity of the simulated damage. Furthermore, the advantages of using transformed DSFs are shown. This is promising for future developments of vibration-based SHM to facilitate improved safety and reliability of WTs at lower costs.

KW - Damage detection

KW - Damage severity estimation

KW - Fukunaga-Koontz transform

KW - Fuzzy computing

KW - Neural networks

KW - Time series methods

KW - Wind turbines

UR - http://www.scopus.com/inward/record.url?scp=84995394428&partnerID=8YFLogxK

M3 - Chapter (peer-reviewed)

VL - 3

SP - 2281

EP - 2290

BT - 8th European Workshop on Structural Health Monitoring, EWSHM 2016

PB - NDT.net

ER -