An evolutionary optimization-based approach for simulation of endurance time load functions

Abstract

In this study, a novel optimization method based on Imperialist Competitive Algorithm (ICA) was proposed to simulate endurance time (ET) excitations. These excitations are monotonically intensifying acceleration time histories that are used as dynamic loading for seismic response analysis and design. The ICA as a population-based evolutionary algorithm was used in this study and the discrete wavelet transform was also employed to reduce the number of optimization variables. Optimal parameter values of the ICA algorithm for simulating ETEFs were evaluated and were utilized to simulate ET excitations. In order to increase the capability of ICA and provide further search in the optimization space, this algorithm was combined with simulated annealing (SA). The new excitation results were compared with the current practice for simulation of ET excitations. It was shown that the proposed ICA-SA method leads to more accurate ET excitations than the classical optimization methods.

Development of an alternative PSO‐based algorithm for simulation of endurance time excitation functions

“…As a result of ETEFs dynamic nature, decision variables of an ETEF simulation problem are highly correlated. In order to consider this correlation in generation of the initial swarm, the method developed by Mashayekhi et al is employed. It is of noteworthy to recall that the method described above is not restricted to solve the specific problem of ETEFs simulation, so it may be readily hired for other problems too.…”

“…Several research studies have been aimed to increase the efficiency of the simulation methods for ETEFs. In the simulation procedure, the objective is to minimize the discrepancy between ETEFs and real ground motions . Thus, equations are expressed in order to account for the discrepancies of ETEFs and real ground motions.…”

“…As mentioned before, simulating ETEFs is an optimization problem that intends to minimize a predefined objective function. The dynamic nature and high number of optimization variables differentiate this problem from other conventional optimization problems . These two issues lead to the presence of many local optima in the problem.…”

“…Although using evolutionary algorithms seem to be justifiable than the classical methods due to the fact that they do not get trapped in local optima, existing ETEFs are yet generated by classical optimization methods . Besides, simulating ETEFs by classical optimization methods is not straightforward and requires several trials and errors to find the best potential solution because optimization results are very sensitive to initial points . Hence, this issue complicates the simulation process of ETEFs.…”

“…These algorithms have been widely used in different engineering practice. In this case, Mashayekhi et al employed the imperialist competitive algorithm (ICA) in simulating ETEFs. Although ICA proposed by Atashpaz‐Garagari has been used to simulate ET excitations, other developed evolutionary algorithms have not been examined to check out whether or not they generate more accurate ETEFs.…”