A probabilistic damage detection approach using vibration-based nondestructive testing

Qindan Huang

doi:10.1016/j.strusafe.2012.01.004

2012-09-01

PRE

Qindan Huang *

Paolo Gardoni

Stefan Hurlebaus

DOI：10.1016/j.strusafe.2012.01.004

原文链接

原文求助

摘要

En 中文

With the aim of improving the accuracy of the assessment of existing structures, damage detection using vibration-based nondestructive testing(NDT) has been extensively studied. It has been recognized that a considerable amount of uncertainties exist in the damage detection process. This paper proposes a novel probabilistic damage detection approach that accounts for the underlying uncertainties. The proposed approach combines two techniques: A Bayesian model updating and a vibration-based damage identification technique (VBDIT). The model updating uses modal frequencies from a damaged structure to build a baseline finite element model (FEM). VBDIT uses mode shapes from the baseline model and the damaged structure to detect damage at local level. The proposed framework makes use of the advantages of the Bayesian model updating and the VBDIT, and compensates for their drawbacks. The sources and types of errors that may occur in the damage detection process are discussed and considered in the proposed formulation. In particular, the proposed approach considers the measurement errors in the vibration tests, the modeling errors in the damage detection process, and the statistical uncertainties in the unknown model parameters. As an application, a finite element model simulating a two-span aluminum beam is used to illustrate the proposed framework. The effects of the measurement and modeling errors on the performance of the proposed damage detection are studied. Modal data can be easily extracted from out-put only responses on an existing structure, making the proposed methodology of practical value. (C) 2012 Elsevier Ltd. All rights reserved.

Keyword:

Nondestructive testing

Damage detection

Vibration test

Modal analysis

Measurement error

Modeling error

Bayesian model updating