Loading...

A Comparative Study on the Application of Metaheuristic Optimization Algorithms For Identification Of Bouc-Wen Hysteresis Model

Sagharichiha, Shahin | 2020

530 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 52700 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Rahimzadeh Rofooei, Fayyaz; Mahdavi, Hossein
  7. Abstract:
  8. In general, modeling errors cause incompatibility between experimental results and the numerical finite element model. To address this issue, the finite element model is being updated to identify the actual dynamic properties of the structure, to evaluate and estimate possible defects of structure, and finally to adapt the behavior of the numerical model and the real structure. As an optimization problem, unknown parameters of the identification model and parameters with uncertainty are being modified. Definition of an appropriate objective function, specifying up-to-date parameters and determining an efficient optimization method are the three main steps of a model-updating problem. On the other hand, in most practical cases, due to the existence of cyclic forces (such as earthquakes) applied to a structure, its members usually depart from the linear area and experience significant deformations in their nonlinear region. In this research work, the Bouc-Wen nonlinear hysteresis model is used as a common tool for modeling structural behavior in the nonlinear region. In order to update the behavior of structure by considering the nonlinear properties, the newly developed water cycle algorithm (WCA), particle swarm optimization algorithm (PSO), imperialist competitive algorithm (ICA) and differential evolution algorithm (DE) are comparatively utilized. Moreover, required programming is conducted in MATLAB software environment, and by using OpenSees software for modeling nonlinear behavior, effective objective functions are enhanced. Comparing the performance of the aforementioned optimization algorithms in terms of computational cost, accuracy of analysis and convergence rate of the responses is one of the main objectives of this research. It is deduced that, by applying various constraints and conditions in dealing with non-linear identification problems, the considered optimizers are gradually faced with more computational difficulties. It is also concluded that, the absolute superiority of the deferential evolution optimization algorithm is due to its internal formulation in achieving to the most reliable results for large scaled optimization problems
  9. Keywords:
  10. Parameters Identification ; Nonlinear Hystersis-Like Behavior ; Optimization Algorithms ; Structural Identification ; Imperialist Competitive Algorithm ; Water Cycle Algorithm

 Digital Object List

 Bookmark

No TOC