基于物联网的物流路径规划与频繁路径挖掘的研究

发布时间：2018-02-13 13:47

本文关键词： 最小代价路径路径规划频繁路径频繁模式数据挖掘　出处：《广西大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着物联网的普及,物联网逐渐应用于众多领域,其中之一是物流领域。在物流领域中路径规划直接影响着物流成本与物流过程中的时间消耗。由于物流中的物品本身带有实时的时空信息,这使得实时的路径规划与从物流数据中挖掘频繁路径成为可能。本文工作主要体现在两个方面。第一,针对基于物联网的物流网络中路径代价的时变性,建立了一种时间依赖的物流网络模型,并在此基础上研究了实时的物流路径规划问题。针对现实中预测准确性随时间推移而降低,以及在预知程度较低时无法获得较好的路径规划等问题,提出了一种带弧代价预知程度参数的最小时间路径算法SWPL以及基于SWPL实时的逐步规划的解决方案。该算法考虑了预测的精度问题,在传统Di jkstra算法的基础上引入了一种与时间相关的弧代价计算方法。实验表明在预知程度较高与在预知程度较低但采用逐步规划的解决方案都能取得良好的路径规划效果。第二,在基于物联网的物流中,会产生海量蕴含时空信息的物品移动数据。这些数据中包含着很多有助于提高物流科学管理的知识,目前从这些数据中找出这些有用的知识采用的主要是数据挖掘技术。其中频繁路径作为反映物流特征的重要知识之一,可为优化物流的路径规划、研究物流的变化规律等提供重要的参考信息。频繁路径的获取是通过频繁序列模式挖掘算法,本文根据物流网络及物流的特征设计了一种充分考虑物流网络拓扑信息的频繁路径序列挖掘算法PMWTI。在该算法中引入了代价容忍度剪枝法,用于候选路径序列的深度剪枝,以去除部分不可能是频繁路径序列的候选路径序列,在一定程度上降低了候选路径序列规模。实验表明,相比没有采用该剪枝方法的同等算法,PMWTI的频繁路径挖掘效率更高。本文工作可为科学的物流管理提供参考,提出的方法可用于物流中实时的路径规划、物流路由的优化、物流规律的发现等。
[Abstract]:With the popularity of the Internet of things, the Internet of things is gradually used in many fields, One of them is the field of logistics. In the field of logistics, path planning has a direct impact on logistics costs and time consumption in the process of logistics. This makes real-time path planning and mining frequent paths from logistics data possible. First, a time-dependent logistics network model is established for the time-variant of path cost in the logistics network based on the Internet of things. On this basis, the real time logistics path planning problem is studied. Aiming at the problem that the prediction accuracy decreases with time in reality, and the better path planning can not be obtained when the degree of prediction is low, and so on. In this paper, a minimum time path algorithm (SWPL) with arc-cost predictive degree parameter and a solution of step by step programming based on SWPL are proposed. The prediction accuracy is considered in this algorithm. Based on the traditional Di jkstra algorithm, a time-dependent arc cost calculation method is introduced. The experimental results show that good path planning results can be achieved in the solutions with higher prediction degree and lower prediction degree but with gradual planning. Second, in Internet-based logistics, there are massive amounts of goods moving data that contain space-time information. These data contain a lot of knowledge that can help improve the scientific management of logistics. At present, data mining technology is the main way to find the useful knowledge from these data, among which frequent paths, as one of the important knowledge reflecting logistics characteristics, can be used to optimize the path planning of logistics. Research on the changing rules of logistics provides important reference information. The acquisition of frequent paths is based on frequent sequential pattern mining algorithm. According to the characteristics of logistics network and logistics, this paper designs a frequent path sequence mining algorithm PMWTI, which fully considers the topological information of logistics network. In this algorithm, the cost tolerance pruning method is introduced for the deep pruning of candidate path sequences. In order to remove part of the candidate path sequence which is impossible to be a candidate path sequence, the size of candidate path sequence is reduced to a certain extent. Experiments show that the efficiency of frequent path mining is higher than that of PMWTI algorithm without this pruning method. This paper can provide reference for scientific logistics management, and the proposed method can be used in real-time path planning, optimization of logistics routing, discovery of logistics rules and so on.
【学位授予单位】：广西大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.44;TN929.5;TP311.13

【参考文献】