A genetic algorithms based procedure for the structural identification of tie-rods

A procedure based on dynamic testing, added mass and Genetic Algorithms (GAs) aimed at the in situ structural characterization of tie-rods is proposed. It is shown that by exploiting only few experimental frequencies of vibration of the tie-rod (principal system S0) along with those of some modified systems (Si with i = 1, 2, · · · , n), that can be simply obtained from the principal system by adding a concentrated mass at a specific position along the tie-rod length, all the tie-rod structural parameters can be identified. In particular, the tie-rod Young’s modulus, the applied tensile load, the rotational stiffness at both restraints as well as the tie-rod material density are identified as the outcomes of an optimization procedure driven by GAs. The fitness function is defined as the discrepancy between the experimental frequencies of vibration of the systems Si and those numerically predicted for a given set of the sought parameters by a finite element based formulation. The value of the parameters is updated into the numerical scheme for the forward computation till the fitness function satisfies the required tolerance. Finally, in order to improve the statistics of the identifications, the Outlier Analysis, in which the discordancy test is carried out via the Mahalanobis distance, is applied to highlight eventual inconsistent identifications that can thus be discarded before the final estimation of the unknown parameters. The procedure has been applied to a metallic tie-rod at the ground floor in Palazzo Poggi (Bologna, Italy), seat of the Rectorate of the University of Bologna.