Algoritmos memeticos aplicados aos problemas de sequenciamento em maquinas

AUTOR(ES)
DATA DE PUBLICAÇÃO

1999

RESUMO

The Single Machine Scheduling Problem (SMS) is one of the most representative in the scheduling area. In this work we initially explore the SMS problem with time constraints (due-dates and setup times). The goal is to minimize the total tardiness, which is characterized by the sum of the delays in the production of all products. The method chosen is based on Memetic Algorithms (MA). MA constitute a class of population metaheuristics that comprise many others, such as Hybrid Genetic Algorithms, Scatter Search, etc. In this thesis the MA implemented is a Genetic Algorithm (GA) with a local search routine that is applied to each new element of the population. Concerning the evolutionary part we study and test several possibilities of recombination operators, mutation, population structures, etc. We also test local search structures that are better fitted to the design of an MA. Performance comparisons are carried out with three different approaches found in the literature. As an addition to this work we also analyze the robustness of the MA and the Fitness Landscape. 80th are extremely important to validate the method and to characterize the difficulty to solve different instances of the problem. As an extension we show some results for the Parallel Machine Scheduling (PMS) problem with sequence-dependent setup times. ...Note: The complete abstract is available with the full electronic digital thesis or dissertations

ASSUNTO(S)

engenharia de produção heuristica algoritmos geneticos

ACESSO AO ARTIGO

Documentos Relacionados