A Comparative Study on the Damage Detection Methods for the Autoregressive Modeling from Ambient Vibration Records
Date Issued
2012
Date
2012
Author(s)
Do, Tien-Quoc
Abstract
Damage detection techniques are the core in structural health moninoring. To obtain damage sensitive feature of a structure, this thesis presents and discusses some techniques using the autoregressive model from ambient vibration records. With the aim of finding accurate modal parameters of structure, this study is focusing on changes of system natural frequencies and damping ratios, which are estimated utilizing AR model. Both singlevariate and multivariate autoregressive models are adopted. For applications in operational modal analysis considering simultaneously the temporal response data of multi-channel measurements, the parameters are estimated by using the least squares method via the implementation of the QR factorization as an essential numerical tool. A new noise rate-based factor called the Noise rate Order Factor (NOF) is introduced for use in the effective selection of model order and noise rate estimation. For the selection of structural modes, an extraction of the system natural frequencies using stability diagram are proposed to identify. This method is thus very effective for identifying the modal parameters in case of ambient vibrations dealing with output-only modal analysis. Accordingly, the first three AR coefficients of the model are used to define the damage sensitive feature because they contain information about natural frequencies and damping ratios. Observations demonstrate effectiveness of damage sensitive feature algorithm as an indicator for damage detection. In addition, an approach based upon the statistical pattern recognition paradigm is assessed. A two-stage prediction model, combining an autoregressive (AR) model and an auto-regressive model with exogenous inputs (ARX), is employed to compute residual error considered as damage sensitive feature.
On the other hand, four different analytic matrices are considered to form Hankel matrix for damage detection: Singular spectrum analysis, SSI-COV, SSI-DATA and Reference-based SSI-COV. Based on the extracted sub-space or null-space from SVD of analytic matrix, damage detection algorithm is developed by considering the orthonormality between the sub-space and null-space. To compare damage indices, the results calculated by this method in chapter 2 are mostly based on Chao & Loh (2012), which will be employed as main reference.
The above mentioned algorithms on structural damage detection are verified using two experimentals: one is a 6-story steel frame with different types of damage scenarios subject to white noise excitation and the other is the bridge scouring test in hydraulic lab.
Subjects
Damage detection
autoregressive model
frequency stability diagram
damage index
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
index.html
Size
23.27 KB
Format
HTML
Checksum
(MD5):435c21aebd52752ad4ddd74f6fb082fc