GA-aided structural identification incorporating artificial boundary conditions

In the subject area of structural identification, the availability of a sufficient amount of valid measurement data is crucial to the success of many theoretically proven algorithms. However, this is often difficult to achieve due to actual structural complications, noises and measurement errors. Among the various response data that may be used for structural identification, the modal frequencies are recognized to be more reliable due to generally good measurement accuracy. In this paper, additional frequency data with artificial boundary conditions are considered together with the natural frequencies for the identification of unknown global structural parameters. A binary genetic algorithm is proposed for the selection of the desired boundary configurations, while a real-coding GA is employed for the actual identification of the structural parameters. The procedure is illustrated using a numerically simulated example.