基于黄金分割遗传算法的VISSIM仿真模型参数校正

摘要

微观交通仿真模型的准度直接决定所模拟交通场景的真实性.提出一种黄金分割遗传算法的VISSIM仿真模型参数校正方法.首先,详细梳理了VISSIM仿真的标准作业流程及模型参数的校正原理;接着,设计了以通过量为指标的目标函数和仿真参数个体的离散化方法;然后,采用黄金分割遗传算法具体设计了VISSIM模型参数的校正方法;最后,以典型城市信控路口为例,分析参数校正后的仿真模型特性,并采用实测路口转向流量数据验证所提出参数校正算法的效用.%The accuracy of microscopic traffic simulation model has direct influence on the reality of the simula-tion scene. By using the golden-ratio genetic algorithm,this paper presented the parameter calibration of VISSIM simulation model. Firstly,the standard process of microscopic traffic simulation and the mechanism of parameter calibration of VISSIM simulation model were discussed in details. Then,based on the throughput index,the ob-jective function and discretization of GA individual parameter were designed,and the calibration algorithm of VISSIM simulation model was developed by golden-ratio genetic algorithm. Finally,taking a typical signalized intersection as the example,it analyzed the characteristics of calibrated traffic flow model and explored the ef-fectiveness of the proposed algorithm by comparing with the observed traffic data in the field. Results show that drivers in China have aggressive driving habits,and the proposed algorithm significantly improve the accuracy of simulation model,while the parameter calibration effectiveness is related to the traffic intensity at the intersec-tion.