Standard Article

Data Preprocessing for Damage Detection

Signal Processing

  1. Andrew Halfpenny

Published Online: 15 SEP 2009

DOI: 10.1002/9780470061626.shm043

Encyclopedia of Structural Health Monitoring

Encyclopedia of Structural Health Monitoring

How to Cite

Halfpenny, A. 2009. Data Preprocessing for Damage Detection. Encyclopedia of Structural Health Monitoring. .

Author Information

  1. nCode International Ltd., Innovation Technology Centre, Rotherham, UK

Publication History

  1. Published Online: 15 SEP 2009


An effective structural health monitoring (SHM) system relies on good quality measurement data. All measured data are susceptible to noise or errors, which are introduced at various stages along the signal path. Some errors gradually develop as the system ages, while others are apparent only under certain operating conditions. In this article, we introduce the most common types of measurement errors and discuss methods for detecting and, where possible, removing them to recover the original measured signal.

The article begins with an overview describing stages along the signal path and the types of errors commonly introduced. We describe the most common configuration errors associated with SHM data acquisition and show how errors can be identified, remedied, or avoided through good system design. The term noise is used to describe an undesirable component within the signal. The article continues by categorizing the types of noise, identifying the cause of noise, and presenting methods of denoising measured data. We conclude by describing techniques that are particularly suited to the analysis of rotating machinery.


  • measurement noise;
  • data anomaly;
  • anomaly detection;
  • anomaly correction;
  • denoising;
  • data cleansing