Analysis of distributed genetic algorithms for solving a strip packing problem

Abstract

This paper presents a solution of a constrained two dimensional strip packing problem using genetic algorithms. The constraint consists of considering three-stage guillotine patterns. This is quite a real constraint motivated by technological considerations in some industries. An analysis of including distributed population ideas and parallelism into the basic genetic algorithm is carried out to solve the problem accurately and efficiently. Experimental evidence in this work shows that the proposed parallel versions of the distributed algorithms outperform their sequential counterparts in time, although there are no significant differences either in the mean best values obtained or in the effort. © 2008 Springer.