A wavelet-based Bayesian damage identification technique using an evolutionary algorithm. Issue 4 (1st December 2016)

• Record Type: Journal Article
• Authors: Varmazyar, Maryam; Haritos, Nicholas; Kirley, Michael
• Is Part Of: Australian journal of structural engineering. Volume 17:Issue 4(2016)
• Page Start: 225
• Page End: 241
• Publication Date: 2016-12-01
• Subjects: Vibration-based structural health monitoring -- wavelet transform -- power spectral density -- Bayesian probabilistic framework -- evolutionary algorithms; Structural engineering -- Periodicals; Structural engineering -- Australia -- Periodicals; Electronic journals; 624.1
• DOI: 10.1080/13287982.2016.1259712
• Languages: English
• ISSNs: 1328-7982

Abstract

Abstract: Structural damage identification is a challenging task, especially when response measurements have local discontinuities and display non-stationarity. This paper presents a one-stage model-based damage identification technique using wavelet power spectra to address this problem. To detect, locate and estimate the severity of damage, the finite element model is updated using a Bayesian probabilistic approach and a covariance matrix adaptation evolution strategy, taking into account the uncertainty caused by measurement noise and modelling error. A range of numerical simulations are used to evaluate the efficacy of the model under different damage scenarios, including: both single and multiple damage locations; varying damage severity; the introduction of noise and modelling errors and incompleteness in the number of captured modes and measurement response data applied to a beam structure. The results obtained across the damage scenarios are observed to be robust. A comparison against existing power spectral density-based methods, which are only applicable in the case of stationary data, indicates that the proposed approach outperforms in almost all damage conditions.