Searching for an efficient method in multiobjective frame optimisation using evolutionary algorithms

Abstract

Elitism has been proved to improve the behaviour of multiobjective evolutionary algorithms. Ways to favour the increment of diversity in the population are recently proposed. Duplicate elimination is showed as an operator that can increase the performance of elitist algorithms. A statistical study of the interaction of this operator with the elitism operator in the non-dominated sorting genetic algorithm is developed. The considered optimisation problem is to minimise the weight of frames with discrete real cross-section types under a multiobjective context: first, the minimization of the constrained weight of the structure, and second, the minimization of the number of different cross-section types. The evaluated test case is a 55 bar sized frame. Statistical results confirm how the introduction of the elitist operator improves significantly the quality of the obtained solution and that the inclusion of the duplicates elimination operator can even outperform those results.