Two meta-heuristics for solving the capacitated vehicle routing problem: the case of the Tunisian Post Office

被引:0
作者
Ines Sbai
Saoussen Krichen
Olfa Limam
机构
[1] Université de Tunis,Institut Supérieur de Gestion de Tunis, LARODEC Laboratory
[2] Université de Tunis El Manar,Institut Supérieur d’Informatique de Tunis, LARODEC Laboratory
来源
Operational Research | 2022年 / 22卷
关键词
Distribution; Post office routing problem; Genetic algorithm; Variable neighborhood search; Hybrid GA-VNS;
D O I
暂无
中图分类号
学科分类号
摘要
Postal sector has a significant role in promoting and improving the services intended for companies and citizens via its various services and its capacity to provide a communication network which ensures rapidity in collecting, transferring and delivering correspondences, funds and goods across the world. Therefore, optimization of the routing system for collection and transport of letters and parcels constitutes an important component of an effective delivery management system. Generally, postal distribution problems are formulated as a Capacitated vehicle routing problem (CVRP) that consists of designing a set of routes, starting and terminating at a central depot and utilize a set of homogenous vehicles to deliver demands to a set of vertices. The objective is to minimize the total transportation cost. Due to its NP-Hardness, we develop in this paper a hybrid metaheuristic that embeds a Variable Neighborhood Search (VNS) in a Genetic Algorithm (GA) in order to accelerate the convergence of the GA to high quality solutions. This combination aims to take advantage of GA’s strength in the exploration and the VNS’s powerful exploitation of the solution space. We propose to include the VNS in the mutation operator of the GA so that the individual space is enlarged and more diversified. Hence, the hybrid algorithm is able to exploit and explore new regions of the search space. The proposed approach is compared to existing methods while applied on benchmark instances. Empirical results driven on five benchmark datasets with a total of 186 instances show that our proposed approach is very competitive in terms of the obtained solutions. Overall, our experiments illustrated that the Hybrid GA-VNS could be a very efficient method for solving the CVRP and its results are comparable with the results of the state-of-the-art. To operationalize our modeling and solution approach, we considered a real case study: the Tunisian Post Office. Results indicate that the proposed HGA-VNS approach improves considerably the solution regarding the existing methods adopted by the Tunisian Post Office.
引用
收藏
页码:507 / 549
页数:42
相关论文
共 159 条
  • [1] Adewumi AO(2018)A survey of recent advances in vehicle routing problems Int J Syst Assur Eng Manag 9 155-172
  • [2] Adeleke OJ(2016)Hybrid large neighbourhood search algorithm for capacitated vehicle routing problem Exp Syst Appl 61 28-38
  • [3] Akpinar S(2017)A variable neighborhood search algorithm for the capacitated vehicle routing problem Electron Notes Discrete Math 58 231-238
  • [4] Amous M(2016)Hybridizations of genetic algorithms and neighborhood search metaheuristics for fuzzy bus terminal location problems Appl Soft Comput 46 220-229
  • [5] Toumi S(2003)A genetic algorithm for the vehicle routing problem Comput Oper Res 30 787-800
  • [6] Jarboui B(2008)An exact algorithm for the vehicle routing problem based on the set partitioning formulation with additional cuts Math Program Ser A 115 351-385
  • [7] Eddaly M(2016)An integration of Lagrangian split and VNS: the case of the capacitated vehicle routing problem Comput Oper Res 78 513-525
  • [8] Babaie-Kafaki S(2016)The vehicle routing problem: state of the art classification and review Comput Ind Eng 99 300-313
  • [9] Ghanbari R(2000)Restructuring of Swiss parcel delivery services OR-Spectrum 22 285-302
  • [10] Mahdavi-Amiri N(2010)Hubbing and routing in postal delivery systems Ann Oper Res 181 109-124