【作者】 陈德良；

【导师】 陈治亚；

【作者基本信息】 中南大学 ， 物流工程， 2010， 博士

【摘要】 现代变化纷繁的激烈市场竞争使得经济系统充满越来越多的不确定性。物流系统作为生产和消费之间的桥梁,是社会经济生活的重要支撑系统,其可靠性也越来越成为企业及社会关注的一个焦点。物流系统任何环节的失效事件都可能导致整个有限承受能力的供应链中断甚至崩溃。物流系统是一个由众多物流单元组成的大系统,它的可靠性程度对企业和社会的平稳运行有着重大的影响。因此,研究物流系统可靠性基础理论具有重要的学术价值和实践意义。本文重点探讨了物流网络可靠性的关键问题及有关应用。首先,探讨物流系统可靠性的定义和计算方法,将可靠性一般理论在物流领域进行扩展。物流系统的可靠性是指整个系统在规定时间和规定条件下,提供的物流服务功能保持在一个规定的允许偏差范围内的概率,它可以根据物流单元结合的串联、并联等逻辑结构分别予以计算。第二,探讨双层物流网络的可靠性特征并进行模拟。以需求地的缺货量为测度,定义了物流网络可靠性,分析了物流网络可靠性的概率特征,并依此构建双层物流网络可靠性模型。模拟分析了不同结构物流网络可靠性,得到了提高物流网络可靠性有三条途径：扩大物流中心的能力、设立更多的配送中心和增加运输配送路径等。第三,探讨物流网络规划的可靠性模型与优化。以物流中心供货服务可靠性为考察目标,建立了基于可靠性和成本的双目标优化的物流中心选址的机会约束规划模型,并针对模型特点,设计了基于随机模拟的遗传算法。第四,探讨库存和路径联合优化问题的可靠性模型与算法。运用机会约束规划方法建立更接近管理实际的库存-路径优化问题数学模型,并综合随机模拟、神经网络和遗传算法设计了求解模型的混合智能优化算法。用实验分析了随机模拟次数对算法的影响。第五,探讨应急物流网络的可靠性优化及替代策略。以应急物资仓库失效时的后备替代策略优化为目标,构建了基于后备替代的应急物流网络的双层规划模型,设计了集成随机模拟技术的遗传算法求解模型。文章最后讨论了提高应急物流网络可靠性的措施。更多还原

【Abstract】 Economic system are more and more full of uncertainty, and in numerous and intense changes in the modern market competition. Being a bridge between production and consumption, logistics system is an important social and economic life support system, and its reliability has increasingly become one focus of business concern. The failure events of any part of the logistics system can lead to the entire supply chain disruption or collapse. Logistics system is a numerous large-scale systems formed by some units, its reliability degree have a major impact on the smooth functioning of business and society. Therefore, the study of logistics system reliability theory has important academic value and practical significance. This article focuses on the concept of the logistics network reliability and calculation, and its optimization based on the reliability of logistics networks.First, this dissertation presents the definition of the logistics system reliability and calculation methods based on the analysis and conclusion of a large number of literatures, and reliability theory will be extended in the logistics fields. Logistics system reliability can be defined as the ability of logistics system providing logistics services in a specified time and specified conditions, which maintaining the allowable deviation in a range of probability. It can be calculated according to the logistics unit combined structures, such as series structures or parallel structures.Second, the dissertation analysizes mainly the characteristics of reliability of bi-level logistics networks, and simulates them. This paper defines logistics networks reliability of measuring the shortage of networks system, and proposes its probability characteristics, and constructs a reliability model of bi-level networks. A simulation and analysis was practiced to demonstrate the reliability of different logistics network structure, and three ways to improve network reliability logistics were obtained, as enhancing the capacity of the logistics center, and setting up more distribution centers and increasing the ability of transportation paths.Third, this dissertation works out the optimization model of the reliability and its algorithms in logistics networks planning. In this paper, the problem of logistics networks planning was mainly addressed, which considering reliability constraints, and an optimization model with Bi-objective was formulated, with which service-reliability-oriented and cost-oriented, and an intelligent algorithm integrated and Stochastic Simulation Methods was applied to solve the model.Fourth, this dissertation addresses the integrated optimization of inventory and routing in the context of Vendor-Managed Inventory, and developes a chance-constrained model and a solution of hybrid intelligent algorithms, of which combining stochastic simulation, artificial neural network and genetic algorithms. An example is finally demonstrated.Fifth, this dissertation studies the alternative strategy of back-up storage, while a failure was happened in emergency supplier, and discusses finally the response measures to improve the reliability of emergency logistics network. A bi-level programming model was constructed, and a hybrid genetic algorithm was designed, which integrated stochastic simulation.更多还原