The main aim of this paper is to present a multimodal model of urban traffic used as predictive model for a traffic control policy. The traffic control policy is based on Model Predictive Control (MPC). The predictive model focus on the private vehicle mode and public transport mode. For the first mode, the model is based essentially on the store-and-forward model. For the second mode, we present an innovative model based on mean behavior of public transport vehicles. Simulation tests show that the model of public transport mode is consistent and worthy to be implemented for a traffic control policy.