EVOLUTIONARY APPROACH FOR MODEL REFINEMENT

In this paper, Genetic Algorithm (GA) based approaches for finite element model topology and parameter adjustment are investigated. Genetic algorithms are one form of directed random search. The form of direction is based on Darwinʹs `survival of the fittestʹ theories. In this paper, an advancement in GA theory is proposed and evaluated to enhance GA search characteristics. The GA-based model refinement process makes use of experimental data to evolve a better correlated finite element model. The model refinement problem is formulated such that the sensitivity of the update to the presence of either analytical or experimental closely spaced modes of vibration is reduced.