A Modified Continuous Reactive Tabu Search for Damage Detection in Beams

The use of vibration-based techniques in damage identification has recently received considerable attention in many engineering disciplines. While various damage indicators have been proposed in the literature, those relying only on changes in the natural frequencies are quite appealing since these quantities can conveniently be acquired. The identification of damage involves an optimization step where response of a continuously updated finite elements model (FEM) is compared with the response of the experimental measurements and error between both responses is minimized. In this paper it is shown that such error function is highly multi-modal and that the same response can be obtained by more than one damage scenario. In order to find these optima a hybrid optimization approach is developed which utilizes two components; namely. Modified Continuous Reactive Tabu Search (MCRTS) and Real Coded Genetic Algorithms. MCRTS, the primary component, is a meta-heuristic capable of finding several optima in a multi-modal search space, which suites the nature of the problem at hand. GAs, the secondary component, although a global optimizer, is used as a local optimizer that is fired in promising regions of the search space as identified by the major component (MCRTS). It is used in favor of direct search methods to account for the presence of minor local optima. In order to test the algorithm, several beams are manufactured and crack damages are induced using wire-cutting, and the natural frequencies are tested experimentally. Such beams have two locations that can give the same response. The developed algorithm managed to find the two sought-for optima consistently in several runs. This proves the merit of this approach as being capable of handling the problem at hand.