24 July 2001 Hybrid global-local identification for structural health monitoring
Author Affiliations +
Genetic algorithms (GA) have been widely used in many optimization and system identification problems. Nevertheless, largely because of the stochastic nature, the approach of using GA has been found to be not efficient in fine-tuning in terms of convergence to the optimal solution from its neighborhood. In this study, the GA approach as a global search tool is combined with a local search (LS) method which is developed on the basis of the classical univariate method, so as to expedite the search process by perturbations near the optimal solution. This LS method, herein named the modified univariate method, does not search all the unknown variables of the problem but only a few selected ones. Furthermore, the initial step size of LS is adjusted according to the average efficiency of each LS operator. For comparison, the Solis-and-Wets LS method is also considered. A numerical example of 10-DOF structure is presented to compare the accuracy and efficiency of the various methods considered.
© (2001) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Chan Ghee Koh, Chan Ghee Koh, C. Y. Liaw, C. Y. Liaw, Y. F. Chen, Y. F. Chen, "Hybrid global-local identification for structural health monitoring", Proc. SPIE 4335, Advanced Nondestructive Evaluation for Structural and Biological Health Monitoring, (24 July 2001); doi: 10.1117/12.434171; https://doi.org/10.1117/12.434171

Back to Top