Generación de Nuevo Conocimiento: Artículos publicados en revistas especializadas - Electrónicos
Comparison of algorithms for reliability-based structural optimization
Fecha
2018-01-03Registro en:
Cortés-Ramos, H. O., Camacho-López, C. J., & Calvo Ocampo, R. A. (2018). Comparison of algorithms for reliability-based structural optimization. Bogotá: doi:10.23967/j.rimni.2017.7.003
Autor
Cortés-Ramos, Henry Octavio
Camacho-López, Carlos Julio
Calvo Ocampo, Rodrigo Andrés
Institución
Resumen
The design of reliable structures requires robust tools that allow the analysis of the
behavior of the system subject to variability in its resistance and applied loads. For this,
there are several formulations and computational algorithms that enable to model the
structural behavior under uncertainties. Under these requirements, the most popular and
reliable optimum design methodology is the reliability based optimization (RBO), which
should be implemented through the combination of high fidelity mathematical or
computational models, e.g. finite element models, efficient and accurate reliability
estimation methods, and efficient and effective engineering optimization algorithms. Most
RBO applications for structural optimization has the latest developments in efficient
computational techniques for simulation and reliability calculations, however, although a
variety of optimization methods exist, they generally do not perform a selection of the
optimization algorithm more appropriate for each application. In this context, the main
contribution of this article is the performing of a comparative study of the computational
performance of optimization algorithms applied in structural optimization by RBO. The
study compared the numerical performance of optimization algorithms in three problems.
Compared algorithms correspond to derivative based algorithms, direct search algorithms,
and bioinspired algorithms; including the most representative algorithms of each category.
The results of the comparative study point out advantages and disadvantages of the use of
the different types of algorithms and allow to conclude on the criteria that must be
considered for the choice of an algorithm that favors the computational performance.