Traffic flow allocation during urban important bridge construction based on mixed graph convolution

Bridge construction often leads to sudden changes in the original traffic network topology, a sharp drop in the capacity of key nodes, and difficulty in accurately allocating traffic flow. Therefore, conducting research on traffic flow allocation during the construction of important urban bridge based on mixed graph convolution. Firstly, by constructing a sandwich structure prediction model and utilizing multi-channel spatial modules to capture static and dynamic spatial dependencies. Secondly, based on the predicted results, a probability model for non-equilibrium traffic flow allocation is established, and the optimal strategy is solved to achieve reasonable flow allocation. Finally, genetic algorithm is used to optimize path selection, while ensuring travel efficiency and alleviating traffic pressure in the construction area. The experimental results show that the path flow allocation balance of the proposed method is stable at 78% -82%, and the probability of choosing a detour path remains at 90.45%-93.56%.