Optimization performance evaluation of evolutionary algorithms: A design problem

Abstract

This paper provides a systematic comparison of four evolutionary optimization algorithms; elitism based genetic algorithm, particle swarm optimization, ant colony optimization and artificial bee colony optimization in terms of their performance with respect to population size, convergence, fitness evaluation and percentage error on an interdisciplinary problem. The case in point is optimized design of high performance concrete mix. The methodology consists of two stages. In the first stage, a huge data base of 450 mix designs garnered through standard research publications were statistically analyzed to elicit upper and lower bounds of certain range constraints and rational ratio constraints of functional parameters. In the second stage, the four algorithms were applied to find the optimized quantities of ingredients constituting the mix. The results indicated that GA was bit high on errors, the other three algorithms showed almost same percentage of error. The convergence of bee colony optimization algorithm was fast followed by particle swarm optimization. © 2011 Springer-Verlag.