文献类型：期刊文献

英文题名：Bilevel Memetic Search Approach to the Soft-Clustered Vehicle Routing Problem

作者：Zhou, Yangming[1,2,3];Kou, Yawen[4];Zhou, MengChu[3,5]

机构：[1]Shanghai Jiao Tong Univ, Data Driven Management Decis Making Lab, Shanghai 200030, Peoples R China;[2]Shanghai Jiao Tong Univ, Sino US Global Logist Inst, Antai Coll Econ & Management, Shanghai 200030, Peoples R China;[3]Macau Univ Sci & Technol, Macau Inst Syst Engn, Macau 999078, Peoples R China;[4]East China Univ Sci & Technol, Dept Comp Sci & Engn, Shanghai 200237, Peoples R China;[5]New Jersey Inst Technol, Dept Elect & Comp Engn, Newark, NJ 07102 USA

年份：2023

卷号：57

期号：3

起止页码：701

外文期刊名：TRANSPORTATION SCIENCE

收录：;EI(收录号：20233014440937);WOS:【SSCI(收录号:WOS:000972886700001)，SCI-EXPANDED(收录号:WOS:000972886700001)】；

基金：Funding: This work was supported by the Macau Young Scholars Program [Grant AM2020011] , Fundo para o Desenvolvimento das Cienciase da Tecnologia (FDCT) [Grant 0047/2021/A1] , the National Natural Science Foundation of China [Grants 61903144, 71871142, and 71931007] , and the Open Project of the Shenzhen Institute of Artificial Intelligence and Robotics for Society [Grant AC01202005002] .

语种：英文

外文关键词：vehicle routing problem; bilevel optimization; heuristic; memetic search; variable neighborhood search

摘要：This work addresses a soft-clustered vehicle routing problem that extends the classical capacitated vehicle routing problem with one additional constraint, that is, customers are partitioned into clusters and all customers of the same cluster must be served by the same vehicle. Its potential applications include parcel delivery in courier companies and freight transportation. Due to its NP-hard nature, solving it is computationally challenging. This paper presents an efficient bilevel memetic search method to do so, which explores search space at both cluster and customer levels. It integrates three distinct modules: a group matching-based crossover (to generate promising offspring solutions), a bilevel hybrid neighborhood search (to perform local optimization), and a tabu-driven population reconstruction strategy (to help the search escape from local optima). Extensive experiments on three sets of 390 widely used public benchmark instances are conducted. The results convincingly demonstrate that the proposed method achieves much better overall performance than state-of-the-art algorithms in terms of both solution quality and computation time. In particular, it is able to find 20 new upper bounds for large-scale instances while matching the best-known upper bounds for all but four of the remaining instances. Ablation studies on three key algorithm modules are also performed to demonstrate the novelty and effectiveness of the proposed ideas and strategies.