Damage detection on a multi-cracked beam by parallel genetic algorithms using modal characteristics

摘要

The present addresses the problem of detecting damage on a multi-cracked reinforced concrete beam. The beam is subjected to a non-symmetrical increasing static load to introduce cracks, after each load step an experimental modal analysis is performed. The crack distribution of the beam is estimated by means of a model based damage detection method based on modal data. A crack is assumed to be directly related to a decrease of stiffness. Optimization is handled by a parallel genetic algorithm. Results show that parallel genetic algorithms give an important increase in performance compared to sequential genetic algorithms. Damage detected with this methodology has a good correspondence with the experimental damage patterns.