Partitioning of urban networks with polycentric congestion pattern for traffic management policies: Identifying protected networks

Large-scale urban networks are usually loaded heterogeneously with a polycentric congestion pattern, resulting in a highly scattered network macroscopic fundamental diagram (NMFD or MFD). Thus, researchers have tried to partition city networks into homogeneous subzones. In this study, a six-step partitioning algorithm is proposed. The framework allows the NMFD information to be used. It combines traffic variables and geographic connectivity to obtain partitions. The framework includes graph definition, data preprocessing, feature handling, clusters and partitions identification, and boundaries reshaping. Tests on a simplified grid network and the city of the Melbourne road network demonstrate the suitability of the framework for characterizing the traffic states by the partitions. The framework on the missing data scenarios and monocentric and polycentric traffic concentrations scenarios, as well as applying multiple data types, has been challenged. Thereafter, parameter impact analysis demonstrates that manipulation of the parameters enables users to find the desired partitions. Last but not least, a comparison with an existing method also implies the uniqueness and efficiency of the developed framework.