| Summary: | 碩士 === 逢甲大學 === 土木工程所 === 96 === This study presents a repetitive sifting based model refinement approach for the health monitoring of structural systems for disaster reduction. The model refinement approach uses the 95% confidence interval of the estimated structural parameters to determine their statistical significance in a least-squares regression setting. When the parameters’ confidence interval covers the zero value, it is statistically sustainable to truncate such parameters. The remaining parameters will repetitively undergo such parameter sifting process for model refinement until all the parameters’ statistical significance cannot be further improved, accounting for possible sampling error and other sources of measurement errors.
The developed model refinement approach is implemented for the power series model to evaluate the structural stiffness. The overall identification methodology from initial model to model refinement for structural stiffness estimation provides reliable indices that are indicative of current conditions of structures for safety assessment, leading to accurate identification as well as controllable design for system vibration control.
This study presents the identification of structural systems under 3-D seismic excitations using a statistically refined Bouc-Wen model. Through limited vibration measurements in the National Center for Research on Earthquake Engineering in Taiwan, the proposed Bouc-Wen model has been statistically and repetitively refined using the 95% confidence interval of the estimated structural parameters so as to determine their statistical significance in a multiple regression setting. The effectiveness of the refined model has been shown considering the effects of the coupled restoring forces in 3-D and of the under -over- parameterization of structural systems. Sifted and estimated parameters such as the stiffness resulting from the methodology developed in this paper can be used for system vibration control.
|
|---|
