BRKGA-QL aplicado ao problema de roteirização e estoques com múltiplos depósitos e entregas fracionadas
Arquivos
Data
2023-01-13
Autores
Volotão, Matheus Santos 
Orientadores
Chaves, Antônio Augusto
Tipo
Trabalho de conclusão de curso
Título da Revista
ISSN da Revista
Título de Volume
Resumo
Neste estudo é abordado o Problema de Roteirização e Estoques com Múltiplos
Depósitos e Entregas Fracionadas (Multi Depot Inventory Routing Problem with Split Deliveries
- MDIRPSD) utilizando o Algoritmo Genético de Chaves Aleatórias (BRKGA) com
algoritmo Q-Learning (BRKGA-QL). O problema é uma variante do Problema de Estoques
e Roteirização (Inventory Routing Problema - IRP), na qual o fornecedor possui múltiplos
depósitos (plantas) e os clientes podem de ser atendido mais de uma vez no mesmo período.
Neste trabalho foi desenvolvido o algoritmo BRKGA-QL com novas heurísticas para
resolver o problema. O algoritmo foi testado para um conjunto contendo 100 instâncias os
quais foram disponibilizadas por Bertazzi et al. (2019) e foram comparadas com o algoritmo
branch-and-cut desenvolvido por Schenekemberg et al. (2021). O BRKGA-QL para
pequenas instâncias conseguiu encontrar soluções relativamente boas durante o processo.
This study addresses the Multi Depot Inventory Routing Problem with Split Deliveries - MDIRPSD using the Random Key Genetic Algorithm (BRKGA) with the QLearning algorithm (BRKGA-QL). The problem is a variant of the Inventory Routing Problem (IRP), in which the supplier has multiple depots (plants) and customers can be served more than once in the same period. In this work, the BRKGA-QL algorithm was developed with new heuristics to solve the problem. The algorithm was tested against a set containing 50 instances which were provided by Bertazzi et al. (2019) and compared with the branch-and-cut algorithm developed by Schenekemberg et al. (2021). BRKGA-QL for small instances managed to find relatively good solutions during the process.
This study addresses the Multi Depot Inventory Routing Problem with Split Deliveries - MDIRPSD using the Random Key Genetic Algorithm (BRKGA) with the QLearning algorithm (BRKGA-QL). The problem is a variant of the Inventory Routing Problem (IRP), in which the supplier has multiple depots (plants) and customers can be served more than once in the same period. In this work, the BRKGA-QL algorithm was developed with new heuristics to solve the problem. The algorithm was tested against a set containing 50 instances which were provided by Bertazzi et al. (2019) and compared with the branch-and-cut algorithm developed by Schenekemberg et al. (2021). BRKGA-QL for small instances managed to find relatively good solutions during the process.