Using sensitivity method and genetic algorithms to model updating

Experimental investigations of actual dynamic properties of civil engineering structures have gained increasing popularity for performance monitoring and maintenance. However, information from such experiments can only be effectively used by correlating the experimental and analytical models. This paper presents a comparison of different model updating techniques: sensitivity method (SM), genetic algorithms (GAs), and a combination of GAs and SM. The SM based updating is routinely used but due to its dependence on the characteristics of search space other global search methods need to be explored. Numerical simulations of a six storey concrete building with assumed dimensions were done using lumped mass models. Different algorithms were applied to update the numerical model. To check the effectiveness of the methods, experiments were also carried out on a 3-storey bookshelf laboratory structure.