Institute of Transportation, MOTC

　　Microscopic traffic models show the basis for driver behavior. The behavior data is quite difficult to collect, especially when large samples are re-quired. Macroscopic traffic models manipulate the data that is easy to collect but lack of the behavior basis of drivers. Mesoscopic traffic models take advan-tage of the above models; mesoscopic models can build the velocity distribution function from the behavior basis and obtain the macroscopic parameters through integration of the distribution function. Mesoscopic models are first proposed by Prigogine and Herman. We construct a new model to relax some poor assumptions of Prigogine and Herman’s – lack of consideration of finite space and instantaneous velocity-changing. Therefore, our model considers a successive slowing-down process and finite baking space. This study also pro-vides the required mathematical definition and the deduction in the proof of model from basic traffic patterns, traffic velocity distribution model, expected velocity distribution model, driver interactive impact model, driver following behavior velocity distribution model, relaxation time and singularity analysis, etc. studies. And we simulate the Prigogine and Hermann model. With these, we might assist researchers to have a better understanding of the model and thus have the ability of further research.