Dynamic partitioning of urban traffic network sub-regions with spatiotemporal evolution of traffic flow.

The spatial partitioning problem of urban traffic network sub-regions has been mainly researched in static mode by considering traffic conditions at a certain time. Nevertheless, traffic flow has strong dynamic characteristics running in urban network; it is necessary to divide the sub-regions according to the correlation of traffic flow in spatiotemporal dimension. Dynamic partitioning is of great significance for capturing the evolution of traffic sub-region and designing the peripheral control strategies in real time. In this paper, a large heterogeneous traffic network is divided into several uniform sub-regions by analyzing the temporal similarity and spatial heterogeneity of the link characteristics to reveal the evolution process of traffic flow. Firstly, a similarity model is defined by the analysis of link autocorrelation based on Moran's I index, and the slow coherency theory combining with α -Cut theory (called α -SC) is implemented to establish the framework of static partitioning. Secondly, a fine-tuning procedure is designed to maintain the homogeneity and compactness of the sub-regions. Finally, the links with high heterogeneity in the sub-regions are identified by heuristic algorithm and then iteratively adjusted, merged and cut to realize the dynamic updating of sub-regions. A case study using the real data set of Farmers Branch city is conducted to verify the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]