Investigação e análise de uma modelagem para o uso do enxame de partículas na otimização do problema de layout de facilidades.

Abstract

This research aimed to investigate a model for the facility layout problem by using particle swarm optmization algorithm, in order to obtain competitive solutions. Facility layout optimization problem discusses allocation of facilities, such as machines, workstations, offices and ordinary departments, in a physical application. Most of the layout problem approaches deals with minimize transport and handling costs, which corresponds from 20 to 50% of operational costs and from 15 to 70% of total manufacturing costs. Particle swarm optimization is a swarm intelligence meta-heuristic which works by exchanging information between individuals from a population, aiming to optimize a specified variable. A survey presented in this work revealed potentiality on solving the facility layout problem using a particle swarm optimization approach. Two double-stage particle swarm optimization methods were developed to adress the problem: first one used slicing trees and second one used the space partitiong method for flexible bay structure. Tem benchmark datasets were used in order to validate methods, and the results were compared with the best outcomes ever published. While first method reached competitive solutions for problems up to ten instances but infeasible solutions for problems bigger or equal to 14 instances, second method was good for problems up to 14 instances, but acceptable for all tested problems.