A multistage FE updating procedure for damage identification in large-scale structures based on multiobjective evolutionary optimization

摘要

This study aims to develop a multistage scheme for damage detection for large structures based on experimental modal data and on finite element (FE) model updating methods applied on simple FE models. In the first stage, occurrence and approximate location of damage is performed by using damage functions in order to decrease the number of parameters to be updated. The goal in the second stage is to identify the specific damaged members and damage extent by considering only the members belonging to the regions detected as damage in the first stage. To improve identification, the optimization procedure is formulated in a multiobjective context solved by using evolutionary algorithms. Modal flexibilities and a damage location criterion dependent on frequencies and mode shapes are used as two objective functions of the multiobjective problem. The proposal is implemented in simulated case studies and in a case study of a real bridge experimentally tested with successful results.