Nearest neighbor and learning vector quantization classification for damage detection using time series analysis

O.R. De Lautour, P. Omenzetter

Research output: Contribution to journalArticle

26 Citations (Scopus)
3 Downloads (Pure)

Abstract

The application of time series analysis methods to structural health monitoring (SHM) is a relatively new but promising approach. This study focuses on the use of statistical pattern recognition techniques to classify damage based on analysis of the time series model coefficients. Autoregressive (AR) models were used to fit the acceleration time histories obtained from two experimental structures; a three-storey laboratory bookshelf structure and the ASCE Phase II experimental SHM benchmark structure in undamaged and various damaged states. The coefficients of the AR models were used as damage sensitive features. Principal component analysis and Sammon mapping were used to firstly obtain two-dimensional projections for quick visualization of clusters among the AR coefficients corresponding to the various damage states, and later for dimensionality reduction of data for automatic damage classifications. Data reduction based on the selection of sensors and AR coefficients was also studied. Two supervised learning algorithms, nearest neighbor classification and learning vector quantization were applied in order to systematically classify damage into states. The results showed both classification techniques were able to successfully classify damage.
Original languageEnglish
Pages (from-to)614-631
Number of pages18
JournalStructural Control and Health Monitoring
Volume17
Issue number6
Early online date25 Mar 2009
DOIs
Publication statusPublished - 1 Oct 2010

Fingerprint

Time series analysis
Damage detection
Vector quantization
Structural health monitoring
Supervised learning
Principal component analysis
Learning algorithms
Pattern recognition
Time series
Data reduction
Visualization
Sensors

Keywords

  • structural health monitoring
  • damage detection
  • time series analysis
  • autoregressive models
  • nearest neighbor classification
  • learning vector quantization

Cite this

Nearest neighbor and learning vector quantization classification for damage detection using time series analysis. / De Lautour, O.R.; Omenzetter, P.

In: Structural Control and Health Monitoring, Vol. 17, No. 6, 01.10.2010, p. 614-631.

Research output: Contribution to journalArticle

@article{b482bb4810534a9aa28f7c47c1886fe5,
title = "Nearest neighbor and learning vector quantization classification for damage detection using time series analysis",
abstract = "The application of time series analysis methods to structural health monitoring (SHM) is a relatively new but promising approach. This study focuses on the use of statistical pattern recognition techniques to classify damage based on analysis of the time series model coefficients. Autoregressive (AR) models were used to fit the acceleration time histories obtained from two experimental structures; a three-storey laboratory bookshelf structure and the ASCE Phase II experimental SHM benchmark structure in undamaged and various damaged states. The coefficients of the AR models were used as damage sensitive features. Principal component analysis and Sammon mapping were used to firstly obtain two-dimensional projections for quick visualization of clusters among the AR coefficients corresponding to the various damage states, and later for dimensionality reduction of data for automatic damage classifications. Data reduction based on the selection of sensors and AR coefficients was also studied. Two supervised learning algorithms, nearest neighbor classification and learning vector quantization were applied in order to systematically classify damage into states. The results showed both classification techniques were able to successfully classify damage.",
keywords = "structural health monitoring, damage detection, time series analysis, autoregressive models , nearest neighbor classification, learning vector quantization",
author = "{De Lautour}, O.R. and P. Omenzetter",
year = "2010",
month = "10",
day = "1",
doi = "10.1002/stc.335",
language = "English",
volume = "17",
pages = "614--631",
journal = "Structural Control and Health Monitoring",
issn = "1545-2255",
publisher = "John Wiley and Sons Ltd",
number = "6",

}

TY - JOUR

T1 - Nearest neighbor and learning vector quantization classification for damage detection using time series analysis

AU - De Lautour, O.R.

AU - Omenzetter, P.

PY - 2010/10/1

Y1 - 2010/10/1

N2 - The application of time series analysis methods to structural health monitoring (SHM) is a relatively new but promising approach. This study focuses on the use of statistical pattern recognition techniques to classify damage based on analysis of the time series model coefficients. Autoregressive (AR) models were used to fit the acceleration time histories obtained from two experimental structures; a three-storey laboratory bookshelf structure and the ASCE Phase II experimental SHM benchmark structure in undamaged and various damaged states. The coefficients of the AR models were used as damage sensitive features. Principal component analysis and Sammon mapping were used to firstly obtain two-dimensional projections for quick visualization of clusters among the AR coefficients corresponding to the various damage states, and later for dimensionality reduction of data for automatic damage classifications. Data reduction based on the selection of sensors and AR coefficients was also studied. Two supervised learning algorithms, nearest neighbor classification and learning vector quantization were applied in order to systematically classify damage into states. The results showed both classification techniques were able to successfully classify damage.

AB - The application of time series analysis methods to structural health monitoring (SHM) is a relatively new but promising approach. This study focuses on the use of statistical pattern recognition techniques to classify damage based on analysis of the time series model coefficients. Autoregressive (AR) models were used to fit the acceleration time histories obtained from two experimental structures; a three-storey laboratory bookshelf structure and the ASCE Phase II experimental SHM benchmark structure in undamaged and various damaged states. The coefficients of the AR models were used as damage sensitive features. Principal component analysis and Sammon mapping were used to firstly obtain two-dimensional projections for quick visualization of clusters among the AR coefficients corresponding to the various damage states, and later for dimensionality reduction of data for automatic damage classifications. Data reduction based on the selection of sensors and AR coefficients was also studied. Two supervised learning algorithms, nearest neighbor classification and learning vector quantization were applied in order to systematically classify damage into states. The results showed both classification techniques were able to successfully classify damage.

KW - structural health monitoring

KW - damage detection

KW - time series analysis

KW - autoregressive models

KW - nearest neighbor classification

KW - learning vector quantization

UR - http://www.scopus.com/inward/record.url?scp=84856361730&partnerID=8YFLogxK

U2 - 10.1002/stc.335

DO - 10.1002/stc.335

M3 - Article

VL - 17

SP - 614

EP - 631

JO - Structural Control and Health Monitoring

JF - Structural Control and Health Monitoring

SN - 1545-2255

IS - 6

ER -