Dynamic Inversion Modeling and Optimization of Branch Road Traffic Flow in Urban Transportation Networks

In the context of the acceleration of urbanization and the rising traffic demand, the accurate deduction of the traffic flow dynamics of unmonitored branch roads is of key significance for the optimization of road network and the formulation of control strategies. This paper constructs a multimodal inversion system based on main road monitoring data, and proposes a differentiated modeling scheme for different road topology and control conditions. The two-way branch coupling model of Y-shaped road network is innovatively proposed, and the high-precision characterization of branch flow evolution law is realized through data preprocessing, dynamic inflection point detection and residual sum of squares minimization parameter optimization. For the multi-branch time-delay coupling scenario, a frequency-domain-time-domain cooperative decoupling method is designed, and a composite model is constructed by combining fast Fourier transform and time-shift compensation, and the parameters are solved by the constrained least squares algorithm. The experiment verifies the high accuracy and reliability of the model, and confirms the feasibility of the multi-source inversion system.