Loading...

Generation of Endurance Time Excitation Functions using Wavelet Transform and Heuristic Optimization Methods

Mashayekhi, Mohammad Reza | 2018

461 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 51365 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Vafai, Abolhassan; Esmaeil Purestekanchi, Homayoon
  7. Abstract:
  8. Endurance Time (ET) method is a dynamic analysis in which structures are subjected to intensifying acceleration time histories. These excitations are also called Endurance Time Excitation Functions (ETEF). The reliability and accuracy of the ET method heavily depends on the accuracy of ETEFs. ETEFs are generated so that ETEFs dynamic characteristics must be consistent with dynamic characteristics of real ground motions. Since the number of ETEFs equations are considerably more than variables noninear optimization is used to generate ETEFs. In order to improve the accuracy of ETEFs two approaches can be used. First, modifying the objective function equations by considering more dynamic characteristics. Second, by applying meta heuristic algorithms and different decomposition methods in simulating ETEFs. In the regard of modifying objective function equations, hysteretic energy of structures and cumulative absolute velocity are included in generating process. CAV is a duration related parameter that highly correlates the potential of a motion to the damage occurred in structures. Duration has not been considered in the generating process of existing ETEFs. It is generally acknowleged that duration of a motion can influence the structural response especially those structures that have degrading and piching behavioyr. In order to include these parameters in generating process, the increasing profile of acceleration time history and increasing variation function of these parameters are determined. In addition, ETEFs simulating equations are developed to used in producing probabilistic excitations that can consider uncertainty in response analysis. In the second approach, different variable definitions of ETEFs are investigated. In the conventional practice, variables are defined in time-domain. In the time domain, acceleration values of ETEFs are considered as optimization variable. In the current thesis, discrete wavelet transform space and increasing sine function space are used to define optimization variables. In the discrete transform space, transform coefficients are decision variables and must be determined during optimization process. Discrete wavelet transform decomposes a singnal into time and frequency components. In the discrete wavelet transform, time variation of a frequency component in time can be detected. This is the main advantage of the wavelet transform in comparison with Fourier transform. Variable definition in discrete wavelet transform space has an advantage that number of considered optimization variables can be reduced. In this space, unnecessary variables can be ignored in the simulation process. In the increasing sine function space, coefficients of base functions are decision variables. In the similar manner to wavelet transform space, optimization variables can be reduced. This variable reduction can help improve the accuracy of results and also reduce analysis computational time. In improving ETEFs accuracy, an efficient optimization algorithm to find optimum values of optimization variables is unavoidable. In this thesis, Imperialist competitive algorithm (ICA) is introduced to simulate ETEFs. Exisiting ETEFs are simulated by using classical optimization methods. ICA is an metha heuiristic optimization method. ICA is inspired from social-political behavior and is based on competition between imperialists
  9. Keywords:
  10. Optimization Algorithms ; Wavelet Transform ; Imperialist Competitive Algorithm ; Endurance Time Method ; Incremental Dynamic Analysis ; Increasing Sine Functions

 Digital Object List

 Bookmark

No TOC