【摘要】 交通状况评价和拥堵预测,是交通管理、调节和诱导的重要依据。通过融合卡口监测数据和出租车为主的浮动车数据,应用数据挖掘的思想构建密度聚类模型,以确定交通状态分类的阈值,从而刻画实际状态;其次建立车辆数-道路流量-道路平均速度关系模型,以预测新增车辆汇入路段后对道路状态的影响。结果表明:多源数据引入交通状态判别后,更有助于精确地划分状态,弥补了单一数据的不足;基于密度的聚类方法,能更有效地刻画不同等级道路的状态。在深圳市的分析验证中结果较可靠,对于助力交通拥堵预判以及缓解、提升智慧交通和智能城市发展意义重大。更多还原

【Abstract】 Traffic condition evaluation and congestion prediction are the basis for traffic management, regulation and guidance. In this paper, the bayonet data and taxi-dominated floating vehicle data are fused, and the density clustering model is constructed using the idea of data mining to determine the threshold of traffic status classification, thereby describing the actual state. Secondly, the vehicle number-flow-average speed relationship model is established to predict the impact of the newly added vehicles on the road state. The results show that the introduction of multi-source data to traffic state discrimination is more helpful to accurately divide the state and makes up for the deficiency of single data. Based on the density clustering method, the state of roads of different levels can be described more effectively. The results of this study are relatively reliable in the analysis and verification of Shenzhen, which is of great significance to help predict traffic congestion and improve the development of smart transportation and smart cities.更多还原