Two-stage damage diagnosis based on the distance between ARMA models and pre-whitening filters

H. Zheng, A. Mita

Research output: Contribution to journalArticlepeer-review

44 Citations (Scopus)


This paper presents a two-stage damage diagnosis strategy for damage detection and localization. Auto-regressive moving-average (ARMA) models are fitted to time series of vibration signals recorded by sensors. In the first stage, a novel damage indicator, which is defined as the distance between ARMA models, is applied to damage detection. This stage can determine the existence of damage in the structure. Such an algorithm uses output only and does not require operator intervention. Therefore it can be embedded in the sensor board of a monitoring network. In the second stage, a pre-whitening filter is used to minimize the cross-correlation of multiple excitations. With this technique, the damage indicator can further identify the damage location and severity when the damage has been detected in the first stage. The proposed methodology is tested using simulation and experimental data. The analysis results clearly illustrate the feasibility of the proposed two-stage damage diagnosis methodology.

Original languageEnglish
Pages (from-to)1829-1836
Number of pages8
JournalSmart Materials and Structures
Issue number5
Publication statusPublished - 2007 Oct 1

ASJC Scopus subject areas

  • Signal Processing
  • Civil and Structural Engineering
  • Atomic and Molecular Physics, and Optics
  • Materials Science(all)
  • Condensed Matter Physics
  • Mechanics of Materials
  • Electrical and Electronic Engineering


Dive into the research topics of 'Two-stage damage diagnosis based on the distance between ARMA models and pre-whitening filters'. Together they form a unique fingerprint.

Cite this