Classification of damage using time series analysis and statistical pattern recognition

O. R. De Lautour, P. Omenzetter

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

1 Citation (Scopus)


The application of time series analysis methods to Structural Health Monitoring (SHM) is a relatively novel and emerging technique. Time series methods are inherently suited to SHM where data is sampled regularly and over a long period of time. This study focuses on the application of statistical pattern recognition techniques to classify damage based on analysis of the time series model coefficients. Autoregressive (AR) models were used to analyse time histories from a structure in both healthy and damaged states. The coefficients of these models were selected as damage sensitive features. Principal Component Analysis (PCA) was used to reduce the dimensionality of the features. Two statistical pattern recognition techniques, Nearest Neighbour (NN) and Learning Vector Quantisation (LVQ) were used to classify damage into states. The results showed that NN classifiers performed well however, improvements could be made using LVQ. The method was applied to a 3-storey bookshelf structure.
Original languageEnglish
Title of host publicationStructural Health Monitoring 2008
Subtitle of host publicationProceedings of the Fourth European Workshop
PublisherDestech Pubns Inc
Number of pages9
ISBN (Print)9781932078947, 1932078940
Publication statusPublished - 23 Jul 2008


  • health
  • pattern recognition
  • principal component analysis
  • structural health monitoring
  • structures (built objects)
  • time series
  • auto-regressive models
  • damage-sensitive features
  • learning vector quantisation
  • nearest neighbours
  • principal components
  • series analysis
  • statistical pattern recognition
  • structural healths
  • time history
  • time series methods
  • time series models
  • time series analysis


Dive into the research topics of 'Classification of damage using time series analysis and statistical pattern recognition'. Together they form a unique fingerprint.

Cite this