Details

Autor(en) / Beteiligte

• Zhang, Guangcai
• Wan, Chunfeng
• Xue, Songtao
• Xie, Liyu

Titel

A global-local hybrid strategy with adaptive space reduction search method for structural health monitoring

Ist Teil von

• Applied mathematical modelling, 2023-09, Vol.121, p.231-251

Ort / Verlag

Elsevier Inc

Erscheinungsjahr

2023

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• •A hybrid global-local strategy for identification of complex and large-scale structures is proposed.•Jaya algorithm is improved by introducing fuzzy clustering, experience learning and Cauchy mutation mechanism.•Sensitivity-based adaptive search space reduction method is developed to redefine the search limits of parameters.•Levenberg-Marquardt, sequential quadratic programming, Nelder-mead simplex method are used as local optimizers.•The performance of proposed methods is validated with numerical and experimental investigations. The difficulty in computational convergence poses challenges of application for traditional heuristic optimization algorithms to solve the optimization-based structural identification problem, especially for the large-scale and complex structural systems where considerable number of unknown parameters and degrees of freedom involved. Unlike the classic identification methods, in this paper, a novel hybrid strategy, coarsely exploring the relatively large search limits with the improved Jaya algorithm and adaptive search space reduction method in the global stage, and then fine-tuning the identified best solution with local optimization methods to the optimum in the local stage, is proposed and evaluated. The improved Jaya algorithm includes three improvements compared to its original version, fuzzy clustering competitive learning, experience learning and Cauchy mutation mechanisms. Gradient based Levenberg-Marquardt method, sequential quadratic programming method and non-gradient based Nelder-Mead simplex method are inserted as local mathematical optimizers to further enhance identification accuracy and efficiency. The superiority of proposed improved Jaya algorithm is validated in optimizing classical and CEC05 benchmark functions by comparing with several state-of-the-art algorithms. Furthermore, the effectiveness of proposed global-local hybrid method is verified by a numerical example of truss structure and an experimental test of the steel grid benchmark structure with incomplete set of noise-polluted measurements. The statistical results show that the improved Jaya algorithm and adaptive search space reduction method combined with sequential quadratic programming can achieve better performance in structural damage identification than other methods. [Display omitted]