Evaluating genetic algorithms with different population structures on a lot sizing and scheduling problem

• Autor(a) principal: Toledo, Claudio Fabiano Motta
• Data de Publicação: 2008
• Outros Autores: França, Paulo Morelato [UNESP], Rosa, Kalianne Almeida
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1145/1363686.1364114; http://hdl.handle.net/11449/70688
• topic: Genetic algorithms; Lot sizing; Multi-population; Scheduling; Soft drink company; Beverages; Binary trees; Computational methods; Diesel engines; Computational results; In lines; Material storages; Population structures; Problem instances; Scheduling problems; Ternary trees; Two types; Scheduling algorithms

This paper studies the use of different population structures in a Genetic Algorithm (GA) applied to lot sizing and scheduling problems. The population approaches are divided into two types: single-population and multi-population. The first type has a non-structured single population. The multi-population type presents non-structured and structured populations organized in binary and ternary trees. Each population approach is tested on lot sizing and scheduling problems found in soft drink companies. These problems have two interdependent levels with decisions concerning raw material storage and soft drink bottling. The challenge is to simultaneously determine the lot sizing and scheduling of raw materials in tanks and products in lines. Computational results are reported allowing determining the better population structure for the set of problem instances evaluated. Copyright 2008 ACM.