Model updating and structural assessment using vibration data with artificial intelligence algorithms
The conventional model updating methods have difficulties in practical applications due to the following three main reasons: 1) high demand on the form and amount of the measurement data, 2) likelihood of being trapped to local optima due to some inherent limitations of the optimization algorithm an...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Theses and Dissertations |
| Language: | English |
| Published: |
2008
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/12039 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | The conventional model updating methods have difficulties in practical applications due to the following three main reasons: 1) high demand on the form and amount of the measurement data, 2) likelihood of being trapped to local optima due to some inherent limitations of the optimization algorithm and, 3) vulnerability to measurement noises. This research programme aims at improving FE model updating techniques in these crucial aspects. The main developments and contributions of this study may be summarized in the following four aspects. 1) A methodology to employ GA in conjunction with the eigensensitivity approach is developed for FE model updating based on a limited amount of modal data; 2) A two-level neural network is proposed to effectively update an FE model involving parameters of different nature, for example stiffness parameters and damping parameters, at two stages; 3) For improved efficiency, a stochastic genetic algorithm (StGA) with a unique stochastic coding scheme is developed in a systematic manner; 4) A GA aided procure for the effective use of artificial boundary condition frequencies (called “ABC” frequencies) in model updating is developed. |
|---|