基于谱聚类算法的城市路网动态分区研究

合理高效的路网分区是交通优化与控制的基础，为适应动态分区的原则并提高分区的效率和划分结果，以谱冈理论为基础结合谱聚类算法进行路网动态分区的研究。分析路网分区的原则并提出路网动态分区的模型框架，根据实时变化的交通流数据和道路交叉口的拓扑结构属性建立谱聚类的相似矩阵，给出结合谱聚类算法的路网分区步骤。通过实际数据验证了算法的实用性。结果表明运用该算法进行路网动态分区可以较好的体现交通流的变化特性，能够得到较为理想的分区结果。

City Transportation Road Network Dynamic Zoning Based on Spectral Clustering Algorithm

Rational and efficient road network zoning is the basis of traffic optimization and control. In order to adapt to the principles of dynamic traffic zoning and improve its efficiency and result, this paper analyzes the dynamic traffic zoning based on spectral clustering algorithm. The principles of road network zoning are analyzed, and a model of road network dynamic zoning is presented. According to the real time data of the continuously changed traffic flow and the property of topology structure of every intersection, a similar matrix of spectral clustering is given. The process of road network zoning based on spectral clustering algorithm is proposed. Finally, the actual traffic flow data prove the practicality of the algorithm. Results show that spectral clustering algorithm provides a useful quantitative research tool for dynamic traffic zoning.