0
  • DE
  • EN
  • FR
  • International Database and Gallery of Structures

Advertisement

Structural damage detection and localization using a hybrid method and artificial intelligence techniques

Author(s):




Medium: journal article
Language(s): English
Published in: Structural Health Monitoring, , n. 5, v. 19
Page(s): 1507-1523
DOI: 10.1177/1475921719887768
Abstract:

In this article, an intelligent scheme for structural damage detection and localization is introduced by implementing a hybrid method using the Hilbert–Huang transform and the wavelet transform. First, the second derivatives of the Discrete Laplacian are computed on Hilbert spectrum parameters at each frequency coordinate, and then, in order to highlight the influence of damage on signals, the data are rescaled and weighted with respect to the variance to adjust the differences in amplitude at different scales. Afterwards, the anti-symmetric extension is applied to deal with the boundary distortion phenomenon. A two-dimensional map is created using the multi two-dimensional discrete wavelet transform. This generates the coefficient matrices of level 2 approximation and horizontal, vertical and diagonal details. Horizontal detail coefficients are used to localize damages due to its sensitiveness to any perturbation. Finally, the validity of the algorithm corresponding to various damage states, the single state damage and multiple state damage, is examined through experimental analysis. The results indicate that the proposed framework can effectively localize cracks on concrete and reinforced concrete beams and can provide reliable crack localization in the presence of noise up to 5% more than the expected noise. In addition, the detection problem is mapped to machine learning tasks (support vector machine, k-nearest neighbours and ensemble methods) to automate the damage detection process. The quality of the models is evaluated and validated using the features extracted from the horizontal detail coefficients. The numerical results show that the ensemble models outperform the other models with respect to accuracy, prediction speed and training time.

Structurae cannot make the full text of this publication available at this time. The full text can be accessed through the publisher via the DOI: 10.1177/1475921719887768.
  • About this
    data sheet
  • Reference-ID
    10562371
  • Published on:
    11/02/2021
  • Last updated on:
    19/02/2021
 
Structurae cooperates with
International Association for Bridge and Structural Engineering (IABSE)
e-mosty Magazine
e-BrIM Magazine