Research on the method of dynamic subarea division of road traffic networks based on floating car data

In large-scale heterogeneous urban road networks, the traffic operation characteristics in different regions exhibit significant disparities. Therefore, it is necessary to formulate corresponding traffic management and control strategies based on the specific features of each region. This research concentrates on partitioning the heterogeneous urban road network into homogeneous sub-areas. By utilizing floating car data (FCD) that encompasses temporal, spatial, and speed information, data processing steps such as outlier elimination, interpolation, and data filtering are implemented to enhance the data accuracy. Subsequently, a similarity index integrating the Pearson correlation coefficient and Euclidean distance is devised to measure the similarity among the traffic flow series of road segments. Incorporating the road network topology structure and similarity matrix into the Gaussian Mixture Model (GMM), a dynamic partitioning algorithm is constructed. Moreover, an innovative dynamic partitioning model is developed, which employs evaluation metrics to determine the optimal number of sub-areas in each period. Through an in-depth analysis of the FCD in Beijing, the experimental results demonstrate that the proposed partitioning algorithm can effectively divide the road network into clusters with high homogeneity. The number and size of the sub-areas change dynamically in accordance with the traffic conditions while ensuring connectivity. The comparative analysis with classical methods (Ncut, Newman) further validates the superiority of the proposed method. In conclusion, this study presents an efficient road network partitioning approach, providing a practical basis for the formulation and evaluation of traffic control strategies.