59 / 2024-06-30 18:08:26

基于高阶依赖关系的轨迹流动态节点重要性预测方法

高阶依赖关系,轨迹流,动态节点重要性预测

二、大数据与复杂系统管理 > 2.复杂系统与高阶网络

摘  要：更深地认识运输系统的内在结构，进行关键节点的识别、预测是社会和军事实际应用中的重要问题。本文旨在解决运输网络中拓扑层上评估方法难以预测真实流量分布以及在负载层上进行流量预测需要大样本数据的问题和现状。基于运输系统中普遍存在的高阶依赖关系，本研究挖掘轨迹流内涵的动力学模型，在粒子尺度上预测节点的重要性。实验结果验证了本文的节点重要性算法对应的排序能够动态地拟合实际负载层上的流量分布，从而为管理者提供新的视角，帮助其更有效地运营运输系统的基础功能。



Abstract: A deeper understanding of the inherent structure of transportation systems and the identification and prediction of vital nodes are critical issues in practical applications for both society and the military. This paper aims to address the challenges and current status in transportation networks, where methods at the topological layer struggle to predict actual traffic distributions and forecasting traffic flow at the load layer requires extensive sample data. Based on the prevalent Higher-order Dependency in transportation systems, this study explores the dynamics of trajectory flows to predict node importance at a small scale. Experimental results validate that the sorting of node importance calculated in this paper dynamically fits real traffic distributions at the load layer, offering managers a fresh perspective to effectively operate the fundamental functions of transportation systems.