【作者】 王保华；

【导师】 何世伟；

【作者基本信息】 北京交通大学 ， 交通运输规划与管理， 2010， 博士

【摘要】 随着我国社会经济的发展以及产业结构的进一步升级,以高附加值产品、行包、邮件为主的快捷货物运输需求总量持续快速增长,同时对服务水平的要求也不断提高。近年来,各种运输方式网络的不断完善为构建一个高效的快捷货物运输系统奠定了坚实基础。在综合运输体系下,根据各种运输方式的技术经济特点,为快捷货物运输合理地配置资源,有利于充分发挥各运输方式的优势、降低快捷货物运输系统的总费用,并最终缓解供需之间的矛盾。本文研究内容主要包括以下几个部分：(1)研究了综合运输体系下快捷货物运输资源配置问题的内涵,包括快捷货物运输供给和需求的特点；国内外快捷货物运输业现状以及我国快捷货物运输业存在的问题；快捷货物运输系统的构成；快捷货物运输资源的定义；快捷货物运输资源配置的含义、原则和影响因素等。(2)研究了综合运输体系下快捷货物运输网络资源利用评价问题。考虑了货物在枢纽停留时间与运输服务频率之间的关系,根据不同运输组织模式下货物在货运场站内的中转作业过程,构建了基于货运场站节点拆分且包含多种运输方式的快捷货物运输服务网络。在网络构建的基础上,基于广义费用构造了考虑运输服务水平的快捷货物运输网络资源利用评价模型,并对联弧费用函数形式对模型的影响进行了分类讨论,给出了分段线性费用函数和非线性费用函数条件下快捷货物运输服务网络的构建方式以及优化模型的形式。给出了基于k短路和Lingo10.0的求解算法,并改进了经典的Double-sweep算法以提高效率。通过若干算例对模型和算法进行了验证,结果表明,文中给出的算法能够获得模型的全局最优解,并可以应用于大规模问题。(3)在研究内容(2)的基础上,研究了综合运输体系下快捷货物运输网络资源配置优化问题。通过引入备选快捷货物运输服务集合,简化模型形式,以最小化运营总费用(包括运输费用和中转作业费用)以及运输网络资源配置费用为目标,构建了快捷货物运输网络资源配置优化模型。通过节点分层与服务区域划分,优化构建备选运输服务集合,从而有效缩减求解空间。给出了四种求解策略,分别是变邻域搜索算法、基于变邻域搜索的模拟退火算法、基于闭合环路的邻域搜索算法、基于k短路和Lingo 10.0的算法。通过若干算例对上述模型和算法进行测试,结果表明,四种求解策略各有其适用条件,应根据问题规模和性质选择合适的方法。在以上研究的基础上,对快捷货物运输网络资源配置优化模型进行了拓展,构建了综合运输体系下快捷货物运输枢纽选址与运输网络资源配置综合优化模型,从而把运输枢纽选址优化问题与运输服务网络设计优化问题结合在一起,并根据模型的性质,给出了一种分阶段的启发式求解策略。(4)在研究内容(3)的基础上,研究了综合运输体系下快捷货物运输网络资源配置与运载工具资源配置动态综合优化问题。通过在快捷货物运输资源配置过程中引入时间因素,构造了包含运载工具利用方案的离散动态运输服务网络,以最小化运营总费用(包括运输费用和中转作业费用)、运输网络资源配置费用以及运载工具资源配置费用为目标构建了动态综合优化模型。给出了一种启发式的求解策略,将综合优化模型分解为配置网络资源和配置运载工具资源的两个子模型,二者之间通过相互反馈,最终达到协同优化的目标。通过服务网络重构的方式,将运输网络资源与运载工具资源配置的动态综合优化模型转化为一类联弧能力也作为决策变量、包含整数变量和连续变量、带时间窗和联弧能力限制的网络设计问题,然后采用基于k短路和优化软件Lingo 10.0的算法对之进行求解。文中还给出若干算例对上述算法进行了验证,结果表明,虽然在快捷货物运输网络资源配置过程中引入时间因素将大大增加网络规模,但本文给出的方法能够在有效时间内得到满意结果,可以满足解决实际问题的需要。(5)在研究内容(3)的基础上,研究了随机环境下快捷货物运输网络资源配置优化问题。考虑快捷货物运输需求的随机性,以最小化运营总费用的期望值为目标,构建了带补偿的快捷货物运输网络资源配置两阶段随机规划模型。针对随机参数分布函数的不同,给出了基于拉格朗日松弛的分解式启发式算法、基于k短路和优化软件Lingo 10.0的算法、基于样本均值逼近的变邻域搜索算法三种求解策略,并对第三种方法的误差进行了估计。给出若干算例对模型和算法进行验证,并着重分析了随机规划模型所得结果的优越性。结果表明,两阶段随机规划模型虽然比确定性规划模型更为复杂,但是能够得到鲁棒性更强的快捷货物运输网络资源配置方案。更多还原

【Abstract】 With the development of society and economy, together with the optimization of industry structure, the demand of express shipment which is composed by high value-added freight and package grows rapidly and continuously. Moreover, the required service level of the demand also increases. These years, the improvement of railroad, road and air network lay a solid fundation for the construction of high-effective express shipment system. Resource planning for express shipment under multi-modal transportation system, according to the characteristic of each mode, is beneficial to making full use of the advantage of each mode, decreasing the total cost of express shipment system and finally solving the contradiction between the supply and demand.The main contents of the paper includes:(1) The implication of express shipment resource planning problem is analysed, including the characteristics of express shipment supply and demand, the current situation of express shipment industry at home and abroad, the composition of express shipment system, and the definition of express shipment resource. Based on this, the definition, principle and influencing factor of express shipment resource planning.(2) The multi-modal express shipment network resource utilization evaluation problem is investigated. The relationship between frequency of express services and delay time at hub yard is firstly taken into consideration in this paper. Then according to the shipment tranfer operation under different policy, the express shipment service network based on hub node partition and multi-modal transportation is constructed.. Based on this network, the evaluation model which takes service level and generalised cost into consideration is proposed. However, the network and the model should be reconstructed according to the form of the cost function (piece-wise linear or non-linear). The solution strategy which is based on k-shortest path and optimization software Lingo 10.0 is presented to solve the model. Several numerical examples are also presented to test the feasibility and efficiency of solution strategy. The testing result shows that the algorithm can obtain the global optimum of the model and can be applied to large-scale problems.(3) Based on Content (2), the multi-modal express shipment network resource planning optimization problem is investigated. By constructing the express freight service set, the model can be simplified. Aiming at minimizing the total operating cost (including transportation cost and delay cost) and resource planning cost, the express shipment network resource planning optimization model is proposed. The searching space is reduced by service area partition and dividing the nodes into several layers. Four algorithms (Variable neighborhood search, variable neighborhood search based on simulated annealing, neighborhood search based on closed-cycles and algorithm based on k-shortest path and Lingo 10.0) are developed to solve the model. Several numerical examples are presented to test the solution algorithms. The testing result shows that four algorithms have different applicable scope. The decision-maker should choose the best algorithm according to the characteristics and scale of the problem.The multi-modal integrated express hub location and network resource planning optimization model is firstly presented in this paper. This integrated model can take the service network design into consideration when investigating hub location problem. According to the characteristics of the model, a multi-phase solution strategy is proposed to solve the integrated model.(4) Based on Content (3), the multi-modal integrated express shipment network resource and fleet resource planning dynamic optimization problem is investigated. Taking time factor into consideration, a discrete dynamic service network which can describe the fleet utilization plan is constructed. Aiming at minimizing the total operating cost (including transportation cost and delay cost) and fleet cost, the dynamic integrated optimization model is proposed. Then, a decomposition heuristic is developed to break up the original problem into two sub-problem, which affect each other when looking for the optimum. A solution strategy which is based on k-shortest path and Lingo 10.0 is also used to solve the model. Finally, several numerical examples are prestented to test the solution algorithms. The testing results show that although time factor increases the size of the network, the solution strategies presented in the paper can obtain good results within tolerable time and is applicable to practical problem.(5) Based on content (3), the multi-modal express shipment network resource planning problem under stochastic environment is investigated. Under the stochastic demand, aiming at obtain a more robust resource plan, a two-stage stochastic programming model with recourse is proposed to minimizing the expected total cost, including resource planning cost, transportation cost and delay cost. According to the distribution function of the stochastic parameters, three solution strategies, which are decompostion heurisitc based on lagrangian relaxation, algorithm based on k-shortest path and Lingo 10.0 and variable neighborhood search based on sample average approximation, are proposed. Since the third strategy uses stochastic simulation to deal with stochastic parameters, the error is also analysed in the paper. The testing results show that although stochastic programming model is more complex than deterministic programming model, it can obtain a more robust resource plan.更多还原