Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation

Simon Hoell, Piotr Omenzetter

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

5 Downloads (Pure)

Abstract

The increasing demand for carbon neutral energy in a challenging economic environment is a driving factor for erecting ever larger wind turbines in harsh environments using novel wind turbine blade (WTBs) designs characterized by high flexibilities and lower buckling capacities. To counteract resulting increasing of operation and maintenance costs, efficient structural health monitoring systems can be employed to prevent dramatic failures and to schedule maintenance actions according to the true structural state. This paper presents a novel methodology for classifying structural damages using vibrational responses from a single sensor. The method is based on statistical classification using Bayes' theorem and an advanced statistic, which allows controlling the performance by varying the number of samples which represent the current state. This is done for multivariate damage sensitive features defined as partial autocorrelation coefficients (PACCs) estimated from vibrational responses and principal component analysis scores from PACCs. Additionally, optimal DSFs are composed not only for damage classification but also for damage detection based on binary statistical hypothesis testing, where features selections are found with a fast forward procedure. The method is applied to laboratory experiments with a small scale WTB with wind-like excitation and non-destructive damage scenarios. The obtained results demonstrate the advantages of the proposed procedure and are promising for future applications of vibration-based structural health monitoring in WTBs.

Original languageEnglish
Title of host publicationSmart Materials and Nondestructive Evaluation for Energy Systems 2017
EditorsNorbert G. Meyendorf
PublisherSPIE
Number of pages12
Volume10171
ISBN (Electronic)9781510608276
DOIs
Publication statusPublished - 19 Apr 2017
EventSmart Materials and Nondestructive Evaluation for Energy Systems 2017 - Portland, United States
Duration: 27 Mar 201728 Mar 2017

Conference

ConferenceSmart Materials and Nondestructive Evaluation for Energy Systems 2017
CountryUnited States
CityPortland
Period27/03/1728/03/17

Fingerprint

Damage Detection
Turbine Blade
turbine blades
wind turbines
Damage detection
Wind Turbine
Instrumentation
Wind turbines
Turbomachine blades
Damage
Partial Autocorrelation
damage
Structural health monitoring
Health Monitoring
Autocorrelation
structural health monitoring
Maintenance
autocorrelation
maintenance
Bayes' Formula

Keywords

  • Damage classification
  • Principal component analysis
  • Statistical classification
  • Time series methods
  • Vibration analysis
  • Wind turbines

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Hoell, S., & Omenzetter, P. (2017). Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation. In N. G. Meyendorf (Ed.), Smart Materials and Nondestructive Evaluation for Energy Systems 2017 (Vol. 10171). [101710D] SPIE. https://doi.org/10.1117/12.2257228

Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation. / Hoell, Simon; Omenzetter, Piotr.

Smart Materials and Nondestructive Evaluation for Energy Systems 2017. ed. / Norbert G. Meyendorf. Vol. 10171 SPIE, 2017. 101710D.

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

Hoell, S & Omenzetter, P 2017, Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation. in NG Meyendorf (ed.), Smart Materials and Nondestructive Evaluation for Energy Systems 2017. vol. 10171, 101710D, SPIE, Smart Materials and Nondestructive Evaluation for Energy Systems 2017, Portland, United States, 27/03/17. https://doi.org/10.1117/12.2257228
Hoell S, Omenzetter P. Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation. In Meyendorf NG, editor, Smart Materials and Nondestructive Evaluation for Energy Systems 2017. Vol. 10171. SPIE. 2017. 101710D https://doi.org/10.1117/12.2257228
Hoell, Simon ; Omenzetter, Piotr. / Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation. Smart Materials and Nondestructive Evaluation for Energy Systems 2017. editor / Norbert G. Meyendorf. Vol. 10171 SPIE, 2017.
@inbook{244de867b5b947b49e6ecf54bfded606,
title = "Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation",
abstract = "The increasing demand for carbon neutral energy in a challenging economic environment is a driving factor for erecting ever larger wind turbines in harsh environments using novel wind turbine blade (WTBs) designs characterized by high flexibilities and lower buckling capacities. To counteract resulting increasing of operation and maintenance costs, efficient structural health monitoring systems can be employed to prevent dramatic failures and to schedule maintenance actions according to the true structural state. This paper presents a novel methodology for classifying structural damages using vibrational responses from a single sensor. The method is based on statistical classification using Bayes' theorem and an advanced statistic, which allows controlling the performance by varying the number of samples which represent the current state. This is done for multivariate damage sensitive features defined as partial autocorrelation coefficients (PACCs) estimated from vibrational responses and principal component analysis scores from PACCs. Additionally, optimal DSFs are composed not only for damage classification but also for damage detection based on binary statistical hypothesis testing, where features selections are found with a fast forward procedure. The method is applied to laboratory experiments with a small scale WTB with wind-like excitation and non-destructive damage scenarios. The obtained results demonstrate the advantages of the proposed procedure and are promising for future applications of vibration-based structural health monitoring in WTBs.",
keywords = "Damage classification, Principal component analysis, Statistical classification, Time series methods, Vibration analysis, Wind turbines",
author = "Simon Hoell and Piotr Omenzetter",
year = "2017",
month = "4",
day = "19",
doi = "10.1117/12.2257228",
language = "English",
volume = "10171",
editor = "Meyendorf, {Norbert G.}",
booktitle = "Smart Materials and Nondestructive Evaluation for Energy Systems 2017",
publisher = "SPIE",
address = "United States",

}

TY - CHAP

T1 - Optimal statistical damage detection and classification in an experimental wind turbine blade using minimum instrumentation

AU - Hoell, Simon

AU - Omenzetter, Piotr

PY - 2017/4/19

Y1 - 2017/4/19

N2 - The increasing demand for carbon neutral energy in a challenging economic environment is a driving factor for erecting ever larger wind turbines in harsh environments using novel wind turbine blade (WTBs) designs characterized by high flexibilities and lower buckling capacities. To counteract resulting increasing of operation and maintenance costs, efficient structural health monitoring systems can be employed to prevent dramatic failures and to schedule maintenance actions according to the true structural state. This paper presents a novel methodology for classifying structural damages using vibrational responses from a single sensor. The method is based on statistical classification using Bayes' theorem and an advanced statistic, which allows controlling the performance by varying the number of samples which represent the current state. This is done for multivariate damage sensitive features defined as partial autocorrelation coefficients (PACCs) estimated from vibrational responses and principal component analysis scores from PACCs. Additionally, optimal DSFs are composed not only for damage classification but also for damage detection based on binary statistical hypothesis testing, where features selections are found with a fast forward procedure. The method is applied to laboratory experiments with a small scale WTB with wind-like excitation and non-destructive damage scenarios. The obtained results demonstrate the advantages of the proposed procedure and are promising for future applications of vibration-based structural health monitoring in WTBs.

AB - The increasing demand for carbon neutral energy in a challenging economic environment is a driving factor for erecting ever larger wind turbines in harsh environments using novel wind turbine blade (WTBs) designs characterized by high flexibilities and lower buckling capacities. To counteract resulting increasing of operation and maintenance costs, efficient structural health monitoring systems can be employed to prevent dramatic failures and to schedule maintenance actions according to the true structural state. This paper presents a novel methodology for classifying structural damages using vibrational responses from a single sensor. The method is based on statistical classification using Bayes' theorem and an advanced statistic, which allows controlling the performance by varying the number of samples which represent the current state. This is done for multivariate damage sensitive features defined as partial autocorrelation coefficients (PACCs) estimated from vibrational responses and principal component analysis scores from PACCs. Additionally, optimal DSFs are composed not only for damage classification but also for damage detection based on binary statistical hypothesis testing, where features selections are found with a fast forward procedure. The method is applied to laboratory experiments with a small scale WTB with wind-like excitation and non-destructive damage scenarios. The obtained results demonstrate the advantages of the proposed procedure and are promising for future applications of vibration-based structural health monitoring in WTBs.

KW - Damage classification

KW - Principal component analysis

KW - Statistical classification

KW - Time series methods

KW - Vibration analysis

KW - Wind turbines

UR - http://www.scopus.com/inward/record.url?scp=85021821176&partnerID=8YFLogxK

U2 - 10.1117/12.2257228

DO - 10.1117/12.2257228

M3 - Chapter (peer-reviewed)

VL - 10171

BT - Smart Materials and Nondestructive Evaluation for Energy Systems 2017

A2 - Meyendorf, Norbert G.

PB - SPIE

ER -