|
|
說明 | 122 p |
附註 | Source: Dissertation Abstracts International, Volume: 70-03, Section: B, page: 1827 |
| Adviser: JUAN M. CAICEDO |
| Thesis (Ph.D.)--University of South Carolina, 2009 |
| Finite element models are idealistic representations of actual structures. They are built to infer the structural behavior or performance under conditions that usually cannot be reproduced in real structures by either physical or monetary constraints. For instance, they can be used to study changes in existing structures, explore innovative designs, or testing the robustness of a control system. When modeling existing structures the numerical model should be updated to better represent the actual structural system through a process called model updating |
| This document presents the development and implementation of an innovative model updating technique. As presented here the model updating methodology is performed based on matching the dynamic characteristics obtained from the real structure. Unlike traditional model updating techniques, the presented technique recognizes that modeling errors, measurements errors, and limited number of sensors might create multiple solutions in the model updating problem. The technique identifies these multiple alternatives and their probabilities based on recorded data from the structure. This dissertation uses Modeling to Generate Alternatives (MGA) to identify physically different alternative solutions and Baye's probability theorem to calculate the probability of each solution |
| The presented model updating methodology was successfully validated using two different structures. First a finite element model of the Emerson Memorial bridge is updated in a deterministic fashion and a family of solutions is found using MGA. Secondly a finite element model of the ASCE Structural Health Monitoring (SHM) Benchmark study structure is updated using data from dynamic and static tests in a probabilistic fashion |
| Results show that the proposed methodology updates the finite element model finding a family of solutions that reduce significantly the difference between the numerical model and the actual structure. The computation of the solutions probability in addition with the objective function value for the solutions is given to the analyst to help him/her to make a final decision on which parameter values to use |
| School code: 0202 |
主題 | Engineering, Civil |
| 0543 |
ISBN/ISSN | 9781109071931 |