服务网络设计用于解决货物运输系统的战术规划问题，广泛应用于交通、物流、生产等领域，涉及到包括大量的车辆、人员等的庞大物流运输网络。由于物流企业的运输网络不断完善，覆盖全国更多的地区，优化物流网络以降低运输服务运营成本变得尤为重要。中国的电商业正处于快速发展之中，消费者的行为受商家促销、节假日等因素的影响导致需求的波动非常大，为快递和零担货物运输服务网络的规划与设计带来了很大的挑战。本论文首先研究考虑多车型的大规模服务网络设计问题，帮助企业决策每个运输服务上使用的各种车型数量，提出相应的数学模型，并设计高效的算法进行求解。结合列生成和割平面算法提供较好的下界，采用局部搜索算法来寻找高质量的可行解。数值实验验证了本文设计的算法的有效性，案例分析证明了考虑多车型能够帮助企业降低物流成本，从而体现了本研究的现实意义所在。为了减小货物运输过程中的风险，尤其是针对危险品货物的运输，同时方便配送和顾客收货，针对服务网络设计问题考虑单路径约束并提出基于环-路径的数学模型。设计了精确算法以提供较好的下界，并采用固定变量法来缩小问题规模，使得问题更容易求解以高效地寻找高质量的可行解，通过数值实验对算法的有效性进行评估。针对考虑不确定需求的服务网络设计问题，采用多面体不确定集来描述需求的不确定性，提出了两阶段鲁棒优化模型。在算法设计上，实现了列与约束生成算法来求解鲁棒优化模型，该算法能够在每次迭代中提供模型的上下界。数值实验表明了列与约束生成算法在求解时间和解质量上都有很好的表现，并分析了鲁棒优化方法的解在结构上的特点，模型比较结果也体现了鲁棒解的优越性。最后为了设计集成服务网络以提高系统中车辆等资源的利用率，针对考虑多种服务类型的服务网络进行优化设计，提出了确定性和两阶段鲁棒优化模型。实现了列与约束生成算法来求解所提出的鲁棒模型，分析了考虑多种服务类型对运输系统的影响，数值实验表明运输系统中服务类型的多样化有助于降低总成本和网络中使用的车辆数。
Service network design is used to address tactical planning issues of the freight transportation system. It is widely applied in transportation, logistics, production, and other fields, and involves large transportation networks including a large number of vehicles and people. As the logistics enterprises keep improving their transportation networks to covers more regions across the country, it is particularly important to optimize the transportation network to reduce the operating costs of transportation services. With the rapid development of China's e-commerce, the behaviors of consumers are affected by the promotion of merchants, holidays and other factors, resulting in great fluctuations in demand, which brings great challenges to the planning and design of express delivery and less-than-truck cargo transportation service networks.This dissertation first studies the large-scale service network design problem considering a heterogeneous fleet, which helps enterprises determine the number of different types of vehicles used in each transportation service. A mixed integer programming mathematical model is constructed for the problem, which is solved efficiently by a hybrid algorithm. The proposed algorithm combines column generation and cutting plane algorithms to provide better lower bounds and employs the local search method to find high-quality feasible solutions. Numerical experiments show the effectiveness of the proposed algorithm. The case study indicates that considering a heterogeneous fleet can help enterprises reduce the total cost, which reflects the practical significance of this research.In order to reduce the transportation risk, especially for the transportation of dangerous goods, and to facilitate the delivery and customer receipt of goods, the single-path constraints are considered for service network design, and a mathematical model based on cycle-path is proposed. An exact algorithm is designed to provide tight lower bounds, and a variable-fixing method is used to reduce the size of the problem, making it easier to solve the problem to find high-quality feasible solutions efficiently. The effectiveness of the algorithm is evaluated through numerical experiments.Regarding service network design considering uncertain demands, the polyhedral uncertainty set is used to describe demand uncertainty, and a two-stage robust optimization formulation is introduced. A column-and-constraint generation algorithm is employed to solve the proposed robust optimization model, which can provide the upper and lower bounds of the model in each iteration. Numerical experiments show that the column-and-constraint generation algorithm has good performance in both computational time and solution quality. The structural characteristics of robust solutions are analyzed, and the results of model comparison also show the superiority of robust solutions.Last, in order to design an integrated service network to improve the utilization rate of resources such as vehicles in the system, both deterministic and two-stage robust optimization models are proposed for service network design considering multiple service types. The column-and-constraint generation algorithm is implemented to solve the proposed robust model, and the impact of considering multiple service types on the transportation system is analyzed. Numerical experiments show that the diversification of service types in the transportation system is helpful to reduce the total cost and the number of vehicles used in the network.
