An approximation to the use of self-adaptive genetic algorithms in weight optimization of 3-D steel trusses

Abstract

In the past few decades structural optimization through metaheuristics has gain recognition in the scientific community; non the less, to guarantee good results we require a good selection of metaheuristic’s parameters. In this paper we propose a multi-chromosome genetic algorithm with self-adaptive parameters, to optimize steel trusses in a three-dimensional space. The design variable are the sections assign to each truss element of the structure. The optimization objective is the minimize the weight of the structure, considering the displacement y maximum stress as constrains. The propose algorithm was applied to the optimization of two trusses, obtaining designs that had a 35% less weight than the initial designs and comparable to results obtained in other papers. However, the adaptation of the parameters allows a more robust optimization process when analyzing different types of trusses and eliminates the initial runs of the optimization algorithm required to calibrate the initial parameters.