Vehicle transport network congestion control in the field of ITS is one of the main means to solve the problem of network congestion. By using a bottom-up agent modeling method, a multi-objective routing decision agent model was constructed to solve the problem. In the model, vehicles were simulated as mobile agents, which determined the routing decision dynamically by a two-objective optimization: the shortest travel distance and the minimum congestion degree of the road network. The multi-objective routing decision model could improve the congestion intersection, and on the other hand, could dig out the common features of congestion intersection to help the network congestion control. The simulation results show that the model can improve road network congestion. And the simulations to compare the effect of network structure with different link densities and link distribution on the road network congestion distribution further indicate that link density of the road network has no effect on the location of the emergent congestion intersections in the network, and the intersection location affects the degree of congestion. It also indicates that the link distribution of the network has a direct effect on the location of the emergent congestion intersections in the network and congestion optimization results.