Analysis of Structural Health Monitoring Data with Correlated Measurement Error by Bayesian System Identification:...

Area of Science:

• Structural Health Monitoring (SHM)
• Data Analysis
• Measurement Error Modeling

Background:

• Measurement errors are critical in SHM data analysis and often exhibit spatial and/or temporal correlations.
• Existing methods fail when simultaneously considering both spatial and temporal correlations in measurement errors.
• A generalized form for spatially and temporally correlated measurement error is needed.

Purpose of the Study:

• To generalize measurement error correlation from spatial or temporal only to spatial-temporal correlation.
• To propose a new form for spatial-temporal correlation and its likelihood function.
• To develop Bayesian system identification for selecting the most suitable measurement error model.

Main Methods:

• Proposed a novel form for spatial-temporal correlated measurement error.
• Constructed multiple candidate model classes: no correlation, spatial, temporal, and spatial-temporal correlation.
• Employed Bayesian system identification to determine posterior probability densities for parameters and model classes.

Main Results:

• Successfully generalized measurement error correlation to include spatial-temporal dependencies.
• Demonstrated the capability of Bayesian system identification in uncertainty quantification at parameter and model levels.
• Validated the approach through applications in model updating and modal frequency prediction under varying conditions.

Conclusions:

• The proposed spatial-temporal correlation model effectively addresses limitations of existing methods.
• Bayesian system identification provides robust uncertainty quantification for SHM.
• Considering correlated measurement error is essential for accurate SHM analysis and prediction.