Road Network Pre-partitioning Method with Priority for Congestion Control

Urban traffic congestion seriously affects the traffic efficiency, causing travel delays and resources wasted directly. In this paper, a road network pre-partitioning method with priority for congestion control is proposed to reduce traffic congestion. Traffic flow feature is extracted based on CNN, and the estimated accuracy of intersection reach 95.32% through CNN-SVM model. Subarea congestion coefficient and intersection merger coefficient are defined to expand the control area of congestion coordination. The association and similarity of intersections are considered using spectral clustering for non-congested intersection partitioning. The results show that the congestion priority control partition method reduces a congestion intersection compared to directly using spectral clustering for subarea partition, and reduces the road network congestion coefficient by 0.05 after 30 minutes than directly using spectral clustering, which is an effective subarea partition method.