Standard Article

Statistical Pattern Recognition

Signal Processing

  1. Hoon Sohn1,
  2. Chang Kook Oh2

Published Online: 15 SEP 2009

DOI: 10.1002/9780470061626.shm054

Encyclopedia of Structural Health Monitoring

Encyclopedia of Structural Health Monitoring

How to Cite

Sohn, H. and Oh, C. K. 2009. Statistical Pattern Recognition. Encyclopedia of Structural Health Monitoring. .

Author Information

  1. 1

    Korea Advanced Institute of Science and Technology, Department of Civil and Environmental Engineering, Daejeon, Republic of Korea

  2. 2

    California Institute of Technology, Department of Civil Engineering, Pasadena, CA, USA

Publication History

  1. Published Online: 15 SEP 2009

Abstract

Structural health monitoring (SHM) problems can often be cast in the context of a statistical pattern recognition paradigm. In this article, a suite of statistical pattern recognition tools applicable to various levels of SHM are introduced. First, multivariate analyses and dimensionality reduction techniques are introduced for effective feature extraction. Here, feature extraction is a step of mining features, which are sensitive to damage of interest, from measured raw signals. Secondly, training data are used to build a statistical model in the inference stage, and the model is used for subsequent decision making for classification and regression problems. If data are available from both undamaged and damaged systems, supervised learning techniques can be used to detect changes in system features extracted from these data that are indicative of damage. If data are only available from the undamaged system, unsupervised learning approaches to subsequent data can be used to identify damage. Finally, ongoing research efforts to address data normalization are presented. Data normalization is a process of separating signal changes caused by operational and environmental variations of the system from those due to structural damage.

Keywords:

  • statistical pattern recognition;
  • structural health monitoring;
  • feature extraction;
  • supervised and unsupervised learning;
  • data normalization