Aplicação de programação geométrica para solução de problemas de estoques com múltiplos objetivos

Abstract

In this dissertation are described the general formulation of the problem multiobjective inventory planning and marketing for Islam (2008) for the case of an a single item and developed a direct generalization of the case with a single item for n items. They are introduced the main inventory models proposed in the literature and in particular, the models EOQ (the Econimic Order Quantity) with backordering allowed and multi-item with restrictions so as to present the main features that de?ne the multiobjective problem of inventory planning and marketing for both a single item proposed by Islam (2008) and for the multi-item. The problems with single item and multi-item besides being multiobjetive are also nonlinear in nature and belonging to the class of problems signomials Geometric Programming. It was necessary to describe and to demonstrate some of the main results of class methods to posteriori Multiobjective Optimization, in particular the method of escalarization Weighted Metrics, used to transform the multiobjective problems into mono-objective problems of Geometric Programming signomiais whose solution was obtained with an algorithm developed for the Condensation technique using Gpposy, belonging to GGPLAB toolbox. After the obtainment of the three formulations for general problems scalarized of inventory planning and marketing for an only item and another three for the case multi-item were obtained particular problems from a set of initial data used so much for the problem formulated with an only item as for the problem multi-item so as to obtain, for three di?erent pairs from associates weights to each objective function, nine problems with a single item and nine problems with ?ve items. The local optimal solutions as well as the viability of constraints for each one of the eighteen problems described were obtained in the chapter of Computational Experiments by means of the algorithm of the Condensation using Gpposy. As Geometric Programmin theory only guarantees solutions of local minimum for problems signomials, the results obtained with escalarization Method of Weighted Metrics guarantees that the solutions of local minimum are locally e?cient solutions so much for the problem multiobjetivo initially proposed for a single item in Islam (2008) how much for your direct generalization multi-item developed.