不同需求条件下流线超网络优化研究

发布时间：2018-04-13 05:20

本文选题：物流需求 + 物流供给能力　；参考：《西南交通大学》2014年博士论文

【摘要】：随着社会经济的不断发展,物流需求越来越复杂,形式上呈现出网络化特征,表现为供应链网络一体化中的物流整合,内容上呈现出多样化特性,表现为固定和随机需求,静态和动态需求。物流需求的复杂性对物流供给服务提出了新的挑战,为了在形式和内容上满足物流需求,应建立与之匹配的物流供给服务网络,使其能及时响应、准确适应物流需求的变化。本文以物流需求与供给为研究视角,以超网络理论、匹配度思想为主要研究依据,结合物流的本质特征,研究流线超网络中需求网络与供给网络的特征与特性,构建了多商品固定需求和多目标固定需求流线超网络优化模型、随机供需流线超网络优化模型及多周期需求流线超网络优化模型。通过优化不同需求特征下流线超网络,解决了物流供给与物流需求的匹配问题,提升了物流服务效率和服务水平,满足了日趋复杂的物流需求。首先,研究了流线超网络理论。从物流需求及供给的角度出发,分析了流线超网络中需求网络和供给网络构成要素,依据物流需求网络特征和供给网络的特性,研究了不同流线超网络的类型。另外,介绍了经典变分不等式及随机变分不等式的定义,分析了变分不等式与其他问题直接的转换关系,研究了变分不等式的几类投影算法,针对随机变分不等式设计了改进随机投影预测算法,并证明其全局收敛性。其次,研究了多商品固定需求流线超网络优化问题。基于固定需求,分析了多商品流通的流线超网络结构,构建了物流需求和供给的费用连续函数,该函数反应了不同商品在不同物流阶段的物流费用。然后,以多商品供需费用匹配最大化为目标建立了优化模型,将模型转换为变分不等式问题,并证明其等价性和解的存在唯一性。最后以煤炭流线超网络为例,采用修正投影算法求解,揭示了物流基础设施能力和商品之间的关系影响供需匹配程度。再次,研究了多目标固定需求流线超网络优化问题。基于固定需求,考虑到物流服务中费用和时间同等重要性,构建了物流费用和时间成本的连续函数,建立了物流供需费用匹配最大和供需时间成本匹配最大的双目标优化模型,将模型转换为相应的变分不等式,证明其等价性和解的存着唯一性,最后以城乡物流流线超网络为例,采用修止投影算法对模型进行求解,算例分析得到物流供给能力和时间等待及滞留是影响供需匹配的较大因素,同时由计算数据表明实行共同配送可以提高物流供需时间成本的匹配程度。进而,研究了随机供需特征流线超网络优化问题。基于终端客户对商品随机需求,构建了物流随机需求网络并分析了其特征,基于物流基础设施的供给能力不确定因素及物流活动会造成商品损失的影响,构建了不确定物流供给网络并分析了其特性。综上研究,建立了随机供需特征的流线超网络优化模型,进一步证明随机优化模型为凸优化模型,将其转换为等价的变分不等式问题,设计改进随机投影预测算法,通过新鲜农产品流线超网络算例对模型进行验证和分析,揭示了针对具体物流需求应提供特定的物流服务,才能避免商品在物流服务中带来损失的结论。最后,研究了多周期需求流线超网络优化问题。考虑到多周期网络中物流线路与物流链之间及物流链与物流序对之间的复杂关系,分析了多周期需求网络和多周期供给网络的特征,建立了多周期物流供需广义费用匹配最大化目标的优化模型,以汽车销售流线超网络为例,求解出了每个周期的供需匹配程度,并根据每个周期的物流需求,给出了针对性的物流供给方案。
[Abstract]:With the continuous development of social economy, the logistics demand is more and more complex, the form of showing the characteristics of the network, is the integration of the supply chain logistics network integration, the content shows diverse characteristics, performance of fixed and random demand, the static and dynamic requirements. Put forward new challenges to the complexity of the logistics demand of logistics supply service in order to meet the demand of logistics, in form and content, should establish the logistics supply and service network and matching, which can accurately and timely response to adapt to the change of logistics demand. According to the logistics supply and demand perspective, with the theory of super network, mainly based on the matching degree of thought, combined with the essential characteristics of logistics. Research on flow characteristics and properties of super network of network and demand supply network, build a multi commodity demand and fixed fixed demand multi-objective optimization model of streamline super network, stochastic demand Streamline super network optimization model and multi period demand flow optimization model super network. By optimizing the different demand characteristics under the flow over the network, to solve the matching problem of logistics supply and demand of logistics, improve the logistics service efficiency and service level, to meet the increasingly complex needs of logistics. Firstly, study the theory of super network. Starting from the line logistics demand and supply point of view, analysis of the super network network and streamline demand supply network elements, according to the characteristics of logistics demand and supply network features of the network, study the different types of flow over the network. In addition, the definition of the classical variational inequalities and stochastic variational inequality, analyzes the conversion relationship. And other problems directly, to study several kinds of projection algorithms for variational inequalities, designed for improved random projection prediction algorithm for stochastic variational inequality, and prove The global convergence. Secondly, this paper studies the optimization problem of super multi commodity network. Based on the fixed fixed demand flow demand, analyses the network structure of multi commodity circulation streamline, construction cost continuous function of logistics demand and supply, the reaction function of different goods in different stages of logistics logistics cost. Then, how to the cost of commodity supply and demand maximization, optimization model is established as the goal, the model is converted into a variational inequality problem, existence and uniqueness and prove its equivalence reconciliation. Finally to streamline the coal super network as an example, using the modified projection algorithm, reveals the relationship between logistics infrastructure and commodity supply and demand matching degree again, on the fixed demand streamline super network multi-objective optimization problem. Based on the fixed demand, taking into account the logistics service cost and time of equal importance, constructs the logistics cost and A continuous function of time cost, set up cost of logistics supply and demand matching double objective optimization model for maximum demand and time cost, the model is converted into the corresponding variational inequality, proving the equivalence and uniqueness of existence, and finally to the urban and rural logistics streamline super network as an example, using the modified projection algorithm is used to solve model, example analysis logistics supply capacity and waiting time and retention is a greater impact factors, at the same time by supply and demand, the calculation data show that the joint distribution can improve the matching degree of logistics supply and demand time cost. Thus, research on the optimization problem of super network stochastic characteristics of supply and demand for goods. Streamline terminal customer demand based on stochastic, construct logistics random demand network and analyzes its characteristics, based on the supply capacity of the logistics infrastructure and logistics activities of uncertain factors will cause the loss of the goods. Ring, we construct uncertain logistics supply network and analyzes its characteristics. In conclusion, a stochastic optimization model of streamline the supply and demand characteristics of super network, further proved that the stochastic optimization model for convex optimization model, and convert it into an equivalent variational inequality problem, improved random projection prediction algorithm, through ultra fresh produce line network examples and analysis of the model, reveals the demand for specific logistics should provide specific logistics services, in order to avoid the loss of goods in the logistics service conclusion. Finally, this paper studies the optimization problem of super network, multi period demand. Considering the streamline logistics lines and logistics chain network and multi cycle logistics chain and logistics order on the complex relationship between the analysis of the characteristics of multi period and multi period supply demand network network, establish a multi cycle logistics supply and demand matching maximum generalized cost The target optimization model is based on the automobile sales streamline super network as an example to solve the matching degree of supply and demand in each cycle, and according to the logistics demand of each cycle, we give a targeted logistics supply plan.

【学位授予单位】：西南交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：F274

【参考文献】