【作者】 葛显龙；

【导师】 王旭；

【作者基本信息】 重庆大学 ， 管理科学与工程， 2011， 博士

【摘要】 随着世界经济的快速发展,物流产业作为国民经济中一个重要的服务产业,正在全球范围迅速发展,并逐渐成为国民经济的基础产业和发展动脉,其发展程度已经成为衡量一个国家现代化程度和综合国力的重要标志。然而,对物流配送的研究还主要停留在传统的配送模式上,随着物联网技术和电子商务的快速发展,物流配送模式已经发生了巨大变化,联合配送、动态配送、大规模跨区域配送等新的配送需求出现,传统的物流配送模式已经难以支持现代物流配送的需求。尤其是通信技术与云计算技术的出现,使得大规模跨区域联合配送成为可能,新的物流配送模式即将诞生。物流配送是现代物流服务供应链中的关键环节,也是开展现代电子商务活动不可缺少的支持部分。对物流配送中车辆调度问题的研究是发展智能交通运输系统、构建综合物流系统和开展电子商务的基础。车辆调度问题经过近五十年的发展研究,己成为运筹学与组合优化领域的研究热点和前沿课题。现实生产和生活中,邮政投递问题、飞机航班安排、铁路车辆编组、码头调运、水运船舶调运、公共汽车调度问题以及电力调度问题等都可以抽象为车辆调度问题。随着电子商务、互联网与通信技术的发展,物流配送与车辆调度问题在各种连锁店、大型商场、快递等领域有广泛的应用前景。因此,对车辆调度问题的深入研究,具有较高的科学意义和工程应用价值。论文的主要研究内容和创新成果如下:①在分析云计算模式与社会化物流配送相互关系的的基础上提出云配送模式对云计算的内涵、发展现状以及应用情况展开分析,结合云计算模式与社会化物流配送的相似特征与联系,提出云配送模式的概念,分析云配送模式的特征及与传统物流配送模式的区别与联系。分析配送资源、配送云服务及配送云的相互关系与作用机理,构建面向社会配送的“公有云”配送服务平台,并对平台构建的关键技术展开分析。②对云配送模式下固定需求的车辆调度问题展开研究对云配送模式下固定需求的车辆调度问题展开研究,构建面向带有里程约束、载重约束和时间窗约束车辆调度问题的数学模型。在分析对比车辆调度问题求解算法的基础上,提出改进遗传算法对多约束条件下的车辆调度问题进行求解,设计针对车辆调度问题的自然数编码的遗传算法,改进传统交叉与变异操作方式,设计种群扩张机制,增强算法的快速寻优和全局收敛能力。结合实验仿真检验模型和算法的有效性,并对比分析不同约束的车辆调度问题求解结果。③对云配送模式下动态需求的车辆调度问题展开研究通过提出时间轴概念,对云配送模式下的动态车辆调度问题展开研究。利用时间轴结合动态信息驱动,记录配送网络发生的信息,将动态车辆调度问题转化为一系列静态车辆调度问题,考虑较为接近实际的约束条件和目标函数,构建具体时刻的车辆调度模型。根据动态车辆调度问题对求解算法时效性要求比较高的特点,提出利用量子理论改进传统遗传算法,设计量子遗传算法,采用量子位多样性的特点设计染色体编码,利用量子门旋转的种群迁移机制提高种群进化效率。结合车辆调度模型设计“初始优化+实时优化”的两阶段求解策略,当动态需求客户提出需求时,利用时间轴标记为不同的时刻,更新配送网络中的信息,实时地进行再优化。结合实验对模型和算法进行了仿真计算,验证算法与模型的有效性。④对云配送模式下联合配送的车辆调度问题展开研究对云配送模式下跨区域联合配送中的关键要素展开分析,提出本文对多配送中心、多车型、开放式动态车辆调度问题处理方法,并根据时间轴概念,建立联合配送的动态车辆调度模型。利用量子理论和云模型理论改进遗传算法,采用量子比特位设计遗传算法染色体编码,利用量子门旋转实现种群进化,采用云模型云滴的稳定性与倾向性特点改进交叉概率和变异概率的设置方式,设计云量子遗传算法,并对算法的性能参数、收敛性及计算复杂度进行分析。结合实验对设计的算法和模型进行实验分析。⑤对智能车辆调度系统的设计与实现展开研究根据云配送模式下配送任务的复杂性与动态性,设计面向对象的车辆调度系统。分析云配送模式下配送任务的需求和系统开发原则,搭建开放式的智能车辆调度系统框架结构。对系统的功能模块进行分析,提出本系统的业务流程和运行结构模式,并展示智能车辆调度系统的主界面和功能模块。最后,对全文研究内容进行总结,展望云配送服务运作模式、车辆调度问题及求解算法的研究前景。更多还原

【Abstract】 With the rapid development of the world’s economic and progress of modern science and technology, as an important national service industry, the logistics industry is developing rapidly worldwide, and has become the basic industries and development artery of the national economy. Its level of development has become an important indicator of a country’s modernization level and overall national strength. However, the logistics researches also stay in the traditional distribution model. With the rapid development of e-commerce and chain stores, logistics distribution model has changed dramatically, appears joint distribution, dynamic distribution, cross-regional large-scale distribution and so new distribution demand, making the traditional logistics model has been difficult to support the needs of modern logistics distribution. Especially the emergence of communication technology and cloud computing, making large-scale cross-regional joint distribution as possible, the new logistics model will be born.Logistics distribution is the key to optimize the modern logistics services supply chain, is also essential support part to expanding modern e-commerce activities. The research of vehicle scheduling problem of logistics distribution is the foundation of developing intelligent transport systems, building integrated logistics systems and expanding e-commerce. Through nearly fifty years’development and study, vehicle scheduling problem has become research hot and cutting-edge topics of operations research and combinatorial optimization field. In realistic production and life, postal delivery problems, flight arrangements, railway vehicles marshalling, terminals transporting, shipping vessels transporting, bus scheduling, power scheduling problem and so can be abstracted for the vehicle scheduling problem. With the development of e-commerce, Internet and communication technology, logistics distribution and vehicle scheduling problem have a wide range of application prospect in the field of chain stores, shopping centers, express and so on. Therefore, the in-depth study of vehicle scheduling problem has high scientific significance and value in engineering.Research contents and innovations of this thesis are as follows:①The cloud distribution model was proposed combining with the relationship between cloud computing model and social logistics distribution.The meaning of cloud computing, development status and application were analyzed, the cloud distribution model was proposed combining with similar characteristics and contraction between cloud computing model and social logistics distribution, the characteristics of cloud distribution patterns and differences and similarities with the traditional logistics model were analyzed. The relationship and action mechanism of distribution resources, delivery cloud services and distribution cloud were analyzed, community-oriented distribution of the "public cloud" service platform was built, and the key technology to building platform was analyzed.②The vehicle scheduling problem with fixed demands under the cloud distribution mode was studied.By having a research on the vehicle scheduling problem with fixed demands under the cloud distribution mode, a mathematical model of vehicle scheduling problem faced with mileage constraint, heavy constraint and time window constraint was established. Based on analysis and comparison the solving algorithms of vehicle scheduling problem, a improved genetic algorithm for solving the vehicle scheduling problem under multi-constraints was proposed, natural number coding genetic algorithm for vehicle scheduling problem was designed to improve the traditional crossover-mutation operation mode, and population expansion mechanism was designed to strengthen the rapid optimization and the global convergence ability. Experimental was combined to test the model and the effectiveness of the algorithm, and the solving results of vehicle scheduling problem with different constraints was contrasted.③The vehicle scheduling problem with dynamic demands under the cloud distribution mode was studied.By proposing the concept of timeline, a research on the vehicle scheduling problem with dynamic demands under the cloud distribution mode was conducted. Timeline and dynamic information driven mechanism was adopted to record the distribution network information, This transforms dynamic vehicle scheduling problem into a series of static vehicle scheduling problems, considering constraint conditions and objective function that more close to the reality, vehicle scheduling model for specific time was established. As vehicle scheduling problem has a high demand in time-sensitive, in this thesis, a method by using quantum theory to improve traditional genetic algorithm was put forwards, designed quantum genetic algorithm, the diversity characteristic of qubits was used to design chromosome coding, and the mechanism of quantum gate spin was adopted to improve the efficiency of the population evolution. Combined with vehicle scheduling model, a“Initial+Real-time”two-stage solution algorithm was design, when requirements proposed by the dynamic demand customer, timeline was used to mark different moments and update the information in the distribution network, then for real-time optimization. Combined with the experimental, simulation computation was used to the designed algorithm to verify the validity of the model and algorithm.④The vehicle scheduling problem of joint distribution under the model of the clouds of distribution was studied.By having an analysis on the key factors of interregional joint distribution under the model of clouds of distribution , the method to solve the Multi Vehicles, multi-distribution centers, open and dynamic vehicle scheduling problem was put forward in this thesis,and according to the timeline concept, a joint distribution of dynamic vehicle scheduling model was build. By using quantum theory and cloud model theory, genetic algorithm was improved, the genetic algorithm chromosome structure was designed by adopting the quantum bits , according to quantum gate rotating population evolution was realized, cloud droplets stability and tendentiousness characteristics of the cloud droplet of cloud model was used to improve the set mode of crossover probability and mutation probability, cloud quantum genetic algorithm was designed and the performance parameters, convergence and computing complexity of the algorithm was analyzed. The designed algorithm and model were got analyzed combined with experiment.⑤The design and implementation of intelligent vehicle scheduling system were studied.According to the complexity and dynamic nature under the model of the cloud distribution, object-oriented vehicles dispatch system was designed. By developing an analysis on principles of system development and requirements of the distributing task under the model of the clouds distribution, a frame structure of open intelligent vehicle scheduling system was build. Through analyzing function module of the system, the business process and operation model of this system was proposed and intelligent vehicle scheduling system of the interface and function modules were on display.Finally, there is a summary on the studies of the whole thesis, the study prospect of operation mechanism of cloud distribution mode, as well as vehicle scheduling problem and its solving algorithm was given.更多还原