联系我们

论文查询结果	返回搜索

论文编号：	16064
作者编号：	2120243792
上传时间：	2026/6/4 21:07:58
中文题目：	AMR辅助拣选与车辆路径的集成优化研究
英文题目：	Integrated Optimization of AMR-Assisted Order Picking and Vehicle Routing
指导老师：	杨静蕾
中文关键字：	拣选与车辆路径；仓配集成优化；订单拣选；自主移动机器人
英文关键字：	Order Picking and Vehicle Routing; Integrated Optimization of Warehousing and Distribution; Order Picking; Autonomous Mobile Robot
中文摘要：	在即时零售业态高速发展、仓储智能化转型持续深化的背景下，城市末端物流对仓配全链路的时效、成本与协同能力提出了更高要求。传统物流运营中，订单拣选与车辆路径规划通常作为两个独立的环节分别优化，割裂了仓储与配送环节的内在耦合关系，导致全链路运营成本高昂、客户服务时效难以保障。与此同时，自主移动机器人（AMR）已成为仓储拣选环节规模化应用的核心装备，为仓配协同优化提供了全新的技术条件，但现有研究鲜有将AMR辅助拣选与车辆路径问题进行深度集成优化，难以适配自动化仓储场景下的仓配一体化决策需求。 针对上述问题，本文聚焦单仓多客户的城市末端物流场景，以AMR辅助的分区拣选模式为基础，构建了AMR辅助订单拣选与带时间窗车辆路径的集成优化模型。模型以仓配全链路总成本最小化为核心目标，综合纳入AMR作业容量与行驶规则、分区拣选串行作业、车辆容量与配送时间窗、暂存区容量等多重现实约束，明确了仓储端批次完成时间与配送端装车时间的双向时序耦合关系，突破了传统割裂式研究的范式局限。 针对模型的NP-hard特性与多约束的求解难点，本文设计了基于三阶段解码机制的自适应大邻域搜索（ALNS）算法。算法采用巨路线表示法统一描述配送序列，提出三阶段解码框架，实现了拣选与配送环节的双向反馈与协同迭代；同时设计了适配集成问题的破坏与修复算子，结合自适应权重更新机制与模拟退火接受准则，有效提升了算法的全局搜索能力与收敛性能。 通过构建不同规模的测试算例集开展多组对比实验，并结合实际企业案例进行应用验证。结果表明，本文设计的ALNS算法对集成问题具备良好的求解性能与收敛稳定性；相较于传统序贯优化方法，集成优化策略能够显著降低仓配全链路总运营成本。案例结果显示，该策略能有效优化AMR配置与作业调度，显著降低延误惩罚成本，充分验证了集成决策模式的优越性。本文的研究成果丰富了自动化仓储场景下仓配一体化优化的理论体系，同时为新零售、生鲜电商等时间敏感型行业的仓配协同运营提供了可落地的决策支持与实践参考。
英文摘要：	In the context of the rapid development of instant retail formats and the deepening transformation of intelligent warehousing, urban last-mile logistics has imposed higher requirements on the timeliness, cost-effectiveness and collaborative capability of the end-to-end warehouse-delivery chain. In traditional logistics operations, order picking and vehicle routing planning are typically optimized as two independent processes. This separation breaks the inherent coupling relationship between warehousing and distribution processes, leading to high operating costs across the entire chain and failure to guarantee customer service timeliness. Meanwhile, Autonomous Mobile Robots (AMRs) have become the core equipment for large-scale deployment in warehouse picking operations, providing brand-new technical support for the collaborative optimization of warehousing and distribution. However, few existing studies have conducted in-depth integrated optimization of AMR-assisted picking and vehicle routing problems, which makes it difficult to meet the integrated decision-making needs of warehouse-distribution operations in automated warehousing scenarios. To address the aforementioned issues, this thesis focuses on the single-warehouse, multi-customer urban last-mile logistics scenario. Based on the AMR-assisted zone picking mode, this thesis constructs an integrated optimization model for AMR-assisted order picking and vehicle routing with time windows. The model aims to minimize the total cost of the warehouse-delivery chain, subject to practical constraints including AMR capacity and routing rules, sequential zone picking, vehicle load limits, delivery time windows, and staging area capacity. It explicitly clarifies the bidirectional temporal coupling relationship between the batch completion time at the warehousing side and the vehicle loading time at the distribution side, thus breaking through the limitations of the traditional decoupled research paradigm. To address the NP-hard nature of the model and the solution difficulties caused by multiple constraints, this thesis designs an Adaptive Large Neighborhood Search (ALNS) algorithm with a three-stage decoding mechanism. The algorithm uses the giant-route representation to uniformly describe delivery sequences and proposes three-stage decoding framework, which realizes bidirectional feedback and collaborative iteration between the picking and delivery processes. It also designs destruction and repair operators adapted to the integrated problem, and combines them with an adaptive weight update mechanism and a simulated annealing acceptance criterion, effectively improving the global search ability and convergence performance of the algorithm. The proposed method is validated by constructing test instance sets of different scales for multiple comparative experiments, and further verified through application cases of real enterprises. The results show that the ALNS algorithm designed in this thesis has good solution performance and convergence stability for the integrated problem. Compared with traditional sequential optimization methods, the integrated optimization strategy can significantly reduce the total operating cost of the entire warehouse-to-delivery chain. Case results further show that this strategy can effectively optimize AMR configuration and operation scheduling, and significantly reduce delay penalty costs, fully verifying the superiority of the integrated decision-making mode. The research results of this thesis enrich the theoretical system of warehouse-distribution integrated optimization in automated warehousing scenarios, and provide feasible decision support and practical references for the collaborative warehouse-distribution operations of time-sensitive industries such as new retail and fresh e-commerce.
查看全文：	预览  下载(下载需要进行登录)