基于NCut算法的城市路网动态分区研究

Research on Dynamic Partitioning of Urban Road Networks Based on NCut Algorithm

摘要: 为改善城市道路交通路网区域拥堵的管控措施，本文提出一种路网动态划分算法。考虑到空间相邻路段的相似性，引入速度时间序列的相关系数去度量变化趋势并构建交通运行特性指标。通过归一化割方法(Normalized Cut, NCut)构建初步路网分区算法，结合自适应调节算法路网进行逐时段的动态分区。利用贵阳市数据验证算法的有效性并评估早晚高峰的分区演变效果。结果表明，所提出的分区算法能揭示不同时期交通演变规律，早晚高峰较平峰有更优的表现，所提出算法能达到有效并理想的分区结果。

Abstract: In order to improve the control measures of regional congestion of urban road traffic network, this paper proposes a road network dynamic classification algorithm. Considering the similarity of adjacent road sections in space, the correlation coefficient of speed time series is introduced to measure the change trend and construct the traffic operation characteristics index. A preliminary road network partitioning algorithm is constructed by the Normalized Cut (NCut) method, which is combined with an adaptive adjustment algorithm for dynamic partitioning of the road network on a time-by-time basis. The effectiveness of the algorithm is verified using data from the city of Guiyang and the effect of the partitioning evolution in the morning and evening peaks is evaluated. The results show that the proposed partitioning algorithm can reveal the traffic evolution pattern in different periods, and the morning and evening peaks have better performance than the flat peaks, and the proposed algorithm can achieve effective and ideal partitioning results.