[论文解读] Search algorithms for efficient logistics chains

Serge Lawrencenko, Irina Albertovna Duborkina|arXiv (Cornell University)|Apr 9, 2015

Process Optimization and Integration参考文献 1被引用 3

一句话总结

本文提出了一种新型物流网络模型，采用效率与不足作为互补度量指标，其中链路效率为各链路效率的乘积，不足为各链路不足的总和。该研究引入了四种搜索算法——两种用于寻找指定节点之间的最高效路径，两种用于确保任意节点对之间的最低服务保障水平——并在能源网络案例研究中通过定量结果展示了损耗最小化的成效。

ABSTRACT

Logistics networks arise whenever there is a transfer of material substance or objects (such as checked baggage on international flights) as well as energy, information, or finance through links (channels). A general concept of logistics network is suggested and motivated for modeling a service of any kind supplied through links between the nodes of the network. The efficiency of a single link is defined to be the ratio of the volume of useful service at the output node to the volume of expended service at the input node of the link (for a specific period of time). Similarly, the efficiency of a chain is the ratio of the volume of service at the output to the volume of service at the input of the chain. The overall efficiency of the chain is calculated as the product of the efficiencies of its links; the more efficiency of the chain, the less are the losses in the chain. This paper introduces the notion of inadequacy of service in such a way that the overall inadequacy of a chain is equal to the sum of the inadequacies of its links. So the efficiencies are being multiplied, whereas the inadequacies are being added. Thus, the antagonistic pair (efficiency, inadequacy) appears to be analogous to the pair (reliability, entropy) in communication theory. Various possible interpretations of the proposed logistic model are presented: energy, material, information and financial networks. Four algorithms are provided for logistics chain search: two algorithms for finding the most effective chain from a specified origin to a specified destination, and two algorithms for finding the guaranteed minimum level of service between any pair of unspecified nodes in a given network. An example is shown as to how one of the algorithms finds the most efficient energy chain from the electrical substation to a specified user in a concrete energy network.

研究动机与目标

• 将物流网络建模为可度量服务效率与不足的链路组合。
• 将整体链路效率定义为各链路效率的乘积，整体不足定义为各链路不足的总和。
• 开发算法以从指定起点到终点识别最高效的物流链。
• 设计算法以确保网络中任意一对节点之间均达到最低保障服务水平。
• 在能源分配网络等实际场景中展示该模型与算法的适用性。

提出的方法

• 将链路效率定义为某一时间段内有用输出服务与输入服务的比值。
• 将链路不足定义为效率的补数，以确保在链路组合中保持可加性。
• 将整体链路效率建模为各链路效率的乘积，以最小化损耗。
• 将整体链路不足建模为各链路不足的总和，以支持可加性优化。
• 开发两种算法，基于效率最大化，用于寻找指定起点与终点之间的最有效链路。
• 开发两种算法，基于不足阈值与全网约束，用于识别所有节点对之间的最低保障服务水平。

实验结果

研究问题

• RQ1 如何通过效率与不足的双重度量来建模物流网络，以反映服务损耗与可靠性？
• RQ2 哪些算法方法能够识别网络中指定两点之间的最高效物流链？
• RQ3 如何确保物流网络中所有可能节点对之间均达到最低保障服务水平？
• RQ4 在链路级性能评估中，基于乘积的效率与基于总和的不足之间存在何种数学关系？
• RQ5 所提出的算法在能源分配等实际物流场景中的表现如何？

主要发现

• 该模型成功地将物流网络建模为类似于通信理论中可靠性和熵的对偶关系，实现了系统化的分析。
• 在具体的能源网络中成功识别出从指定起点到终点的最高效链路，证明了其实际适用性。
• 最低保障服务水平算法确保网络中任意一对节点的服务水平均不低于预设阈值。
• 采用可加的不足与可乘的效率，使得在复杂链路中实现可扩展且数学一致的优化成为可能。
• 在变电站网络的案例研究中证实，该算法能有效最小化损耗并最大化服务交付效率。