FE model updating using artificial boundary conditions with genetic algorithms

Zhenguo Tu, Yong Lu

Research output: Contribution to journalArticlepeer-review

Abstract / Description of output

In the realm of finite element (FE) model updating and damage identification, an outstanding issue is with the limited amount of reliable response data that may be used to perform an inverse procedure. This problem can restrict the number or types of physical parameters that may be identified or updated, and it could also result in an erroneous identification of the parameters due to insufficient sensitivity of the data set. To tackle this problem, an effective enlargement of the data set is desired. This paper presents a genetic algorithm (GA)-based methodology to make effective use of the artificial boundary condition (ABC) frequencies for FE model updating. The ABC frequencies can be obtained through the measurement of the incomplete frequency response functions of the structural system with a limited number of sensors, and thus they can be of similar measurement accuracy as the natural frequencies. In the present methodology, a binary coding GA is proposed for the selection of the desired artificial boundary conditions; while for the actual updating of the FE model, a procedure based on a real coding GA is implemented. Numerical examples are provided to demonstrate the effectiveness of the proposed approach in the FE model updating. (C) 2007 Elsevier Ltd. All rights reserved.

Original languageEnglish
Pages (from-to)714-727
Number of pages14
JournalComputers and Structures
Volume86
Issue number7-8
DOIs
Publication statusPublished - Apr 2008

Keywords / Materials (for Non-textual outputs)

  • finite element model updating
  • modal frequencies
  • artificial boundary
  • omitted coordinate set system
  • sensitivity analysis
  • genetic algorithm (GA)
  • DAMAGE DETECTION
  • IDENTIFICATION
  • FREQUENCIES
  • SELECTION
  • LOCATION

Fingerprint

Dive into the research topics of 'FE model updating using artificial boundary conditions with genetic algorithms'. Together they form a unique fingerprint.

Cite this