With the rapid development of the city, private car ownership is increasing year by year, traffic congestion problem has aroused great concern to the relevant departments in many cities. The microscopic traffic simulation software is a powerful tool to evaluate traffic solution, so the accuracy of the model is very important. At present, the traffic simulation software used by the most of our research institutions are imported from abroad, so carrying out model parameters calibration according to the actual traffic operation condition in our country on the software model parameters calibration is the prerequisite and foundation to perform other work. The common algorithm of model parameter calibration is genetic algorithm, but the genetic algorithm will cost a lot of time in the iterative process. At the same time, most of the parameter calibration methods are separate programs, and have not realized the parameter correction automatically. Based on the above two issues, the thesis has implemented a new method of automatically parameter calibration program for VISSIM. In this thesis, following work has been finished: Firstly， the thesis provides an overview of microscopic traffic simulation, and summarized the current research status. The research orientation for parameter calibration is along two main lines, and the focus of this thesis is on the research for the model calibration algorithm. In this thesis, the core parameters of car following model and lane-changing of VISSIM are introduced in detail, and then, the method of choosing evaluation index and parameter for calibration were analyzed. Secondly, the thesis take genetic algorithm as parameter calibration method, and using the trained generalized regression neural network to predict the output of VISSIM , that can avoid the waste of time in the iterative process of genetic algorithm. After that, a case study of the Zhong Guan Street intersection in Beijing is carried out, the above parameter calibration method is verified. The results show that, this method can effectively improve the ef...
修改评论