【作者】 彭扬；

【导师】 陈华平；

【作者基本信息】 中国科学技术大学 ， 管理科学与工程， 2007， 博士

【摘要】 物流配送网络设计与优化是物流系统领域一个重要的决策问题，一般可以分为三个设计层次：战略层、战术层和运作层，三个层次之间既有各自独立的决策问题又由于相互影响所以存在所谓的集成设计问题。解析方法是解决类似优化问题的常用方法，本文首先对物流配送网络设计进行系统分析和国内外研究现状进行回顾和综述，就网络设计中的一些重要课题进行数学模型构建和算法求解研究。具体内容包括：1．针对二级网络中的协作分销设计问题进行研究，探讨多供应点、多产品、多需求点单时段的物流网络协作供应为研究对象，以供需产品的类型差异来驱动多个优化目标间的平衡，并设计了改进的遗传算法来求解这个复杂的组合优化问题。2．为了求解多级物流网络设计这样一个NP-Hard问题，给出了两层嵌套的混合遗传算法，在算法流程中引入了模拟退火算法以改进遗传算法的性能。同时也探讨了混合算法的一般原则，并通过算例仿真分析所提出算法的应用和有效性。3．提出了基于粒子群优化算法(PSO)的求解定位一运输安排问题(LRP)的方法，该方法采用两层求解机制，上一层应用PSO方法设计和优化定位分配问题，并在计算粒子适应值时应用下层的车辆路径计划结果，并采用了新颖的粒子编码和算法机制。该方法的提出对于LRP问题的求解是一个创新的有效的尝试。4．进一步深入探讨了LRP问题的形式化模型表达和基于PSO的求解方法。给出了基于图论的数学模型，并为求解一类特殊的LRP问题，提出改进的离散粒子群优化算法，该方法采用整体优化的思想，将LAP和VRP集成在一起。通过合适的粒子编码方式，并改进粒子的运动方程，引入相应的算子改善了算法的适用性和性能。通过仿真实验证明了该算法的有效性文章最后通过一个有关浙江电信物流网络规划的实践研究项目，对现实中的物流网络设计问题进行了系统性的分析和解决。更多还原

【Abstract】 Logistics network design and optimization is an important problem for decision-making in the logistics system, which usually can be divided into three layer problem for design: strategy layer, tactics layer and operation layer. They not only each have own problems independently, but also have the integrated problem because of the mu tual influence in them. Analytic method is the common way for the similar optimization problems, the paper firstly gave the problem a systems analysis and literature reviewed on the domestic and international researchers. And then studied on the model and solution for some important problems in LNDP, This main results obtained in this dissertation may be summarized as follows:1. Research on an important problem in the two-echelon logistics networks, to tackle the multiple supply depots, multiple kind products, multiple demands and single period collaborated distribution problem, balanced and transformed the multi-objective problem by diversity of products, and design an improved genetic algorithm to solve the complex combination optimization problem;2. Presented a two layer nested genetic algorithm for multi-layer logistics networks design, which was mixed with simulation annealing algorithm in the method for improved the algorithm’s performance. And also discuss on the common principles of algorithm mix methodology, then gave an solution example of the problem, which showed the application and availability of the presented method;3. Location-Routing Problem(LRP) is a NP-hard problem in the integrated logistics system, a solution method based on PSO Problem is presented, which adopted two-layer solution mechanism, the upper layer and lower layer both applied PSO but different code of particle to solve LAP and VRP respectively, and the LAP layer should call the VRP layer’s solution to evaluate the particle’s fitness. The method is a innovative solution for LRP, and Simulation results of designing LRP examples demonstrated its’ effectiveness.4. An mathematical model for LRP based on network theory is put forward, and then for tackling a kind of special LRP, an improved discrete Particle Swarm Optimization algorithm is presented Which adopted the principle of whole optimization, integrated the LAP and VRP, introduced a appropriate coding for the particle, improved the particle’s move equation, also added the mutation operator and disturbing operator against the population identical tendency, the applicability and performance of the algorithm has improved. At the last part of the paper, discuss a practical research project about the Zhejang telecom company logistics distribution networks planning, gave the reality problem of LNDP a systems analysis and successful settled.更多还原