Standard Article

Probabilistic Approaches to Sensor Layout Design, Data Processing, and Damage Detection

Simulation

Probabilistic Approaches

  1. Sankaran Mahadevan,
  2. Xiaomo Jiang,
  3. Robert F. Guratzsch

Published Online: 15 SEP 2009

DOI: 10.1002/9780470061626.shm072

Encyclopedia of Structural Health Monitoring

Encyclopedia of Structural Health Monitoring

How to Cite

Mahadevan, S., Jiang, X. and Guratzsch, R. F. 2009. Probabilistic Approaches to Sensor Layout Design, Data Processing, and Damage Detection. Encyclopedia of Structural Health Monitoring. .

Author Information

  1. Vanderbilt University, Nashville, TN, USA

Publication History

  1. Published Online: 15 SEP 2009

Abstract

This article presents probabilistic methodologies toward the design of a reliable structural health monitoring system, by addressing several types of uncertainties in sensor performance and data, including sensor layout effects and sensor data noise, incompleteness, and variability. The sensor layout is designed to maximize the reliability of damage detection by combining probabilistic finite element analysis, damage detection algorithms, and reliability-based optimization techniques. A Bayesian discrete wavelet packet transform-based denoising approach is presented to perform data cleansing prior to damage detection. A nonparametric system identification method is applied when using incomplete sensor data. A Bayesian method is presented to incorporate possible uncertainties in both sensor data and model prediction and to provide a quantitative measure of confidence in structural condition assessment. These methodologies are illustrated for application to an aerospace structure thermal protection system panel and to a five-story building frame structure, representing two different disciplines.

Keywords:

  • structural health monitoring;
  • sensor layout;
  • damage detection;
  • reliability;
  • Bayesian statistics;
  • wavelets;
  • denoising;
  • thermal protection system