无信号交叉口仿真模型参数标定优化与验证  

Optimization and Verification of Calibrating Parameters of Simulation Model of Unsignalized Intersection

作　　者：刘博[1] 周晨静[1] 林建新[1] LIU Bo;ZHOU Chen-jing;LIN Jian-xin(Beijing Future Urban Design Advanced Innovation Center,Beijing University of Civil Engineering and Architecture,Beijing 102616,China)

机构地区：[1]北京建筑大学北京未来城市设计高精尖创新中心,北京102616

出　　处：《公路交通科技》2021年第10期129-136,共8页Journal of Highway and Transportation Research and Development

基　　金：国家自然科学基金项目(5170080357)。

摘　　要：为优化与验证微观交通仿真模型的微观参数标定方法,结合已有研究针对仿真模型全局参数与局部参数的讨论及对标定结果不同取值方式进行交叉组合,形成4种新的微观仿真模型参数标定方法,并选取等权无信号交叉口作为仿真场景进行分析验证。基于VISSIM仿真平台建立微观仿真模型,以通行能力作为校核指标,敏感性分析选取6个对校核指标影响较大的微观仿真模型参数作为局部参数标定对象,并通过遗传算法进行优化迭代。选取偏差在4%以内的参数组合方案作为标定结果取值分析对象,针对标定结果聚类递归取值选取分散度作为描述参数集中程度的指标。实测交叉口交通数据进行交叉验证试验,对比了4种参数标定方法模拟校核指标的精度,分析了不同全局参数标定方法以及标定结果不同取值方式对微观参数最终标定结果的影响。结果表明:工程实测获取全局参数标定的仿真模型更符合城市车辆实际运行特性,标定结果聚类递归取值能有效提升模型参数标定结果的可移植性;实测全局参数结合聚类递归获取局部参数的方法最为有效,可作为微观交通仿真模型参数标定方法优化研究方向,且为无信号交叉口微观交通仿真模型构建提供基础标定参数。To optimize and verify the microscopic parameter calibration method of traffic simulation model,combining the existing researches on the discussion of the global and local parameters of the simulation model and the cross-combination of different values of the calibration result,4 new micro-simulation model parameter calibration methods are formed,and the equal right unsignalized intersection is selected as the simulation scene for analysis and verification.The microscopic simulation model is established based on the VISSIM simulation platform,selecting traffic capacity as the check indicator,6 micro-simulation model parameters that have a greater influence on the check indicator as the local parameter calibration objects for sensitivity analysis,the optimal iteration is conducted by genetic algorithm.Selecting the parameter combination scheme with deviation of less than 4%as the value analysis object of the calibration result,and selecting dispersion degree as the indicator to describe the parameter concentration degree according to the clustering recursive value of the calibration result,the intersection traffic data are measured and a crossverification experiment is carried out,the accuracies of the calibration indicators simulated by the 4 parameter calibration methods is compared,and the influences of different global parameter calibration methods and different valuing methods of the calibration result on the final calibration result of the micro parameters are analyzed.The result shows that(1)simulation model of global parameter calibration obtained from engineering measurement is more consistent with the actual operation characteristics of urban vehicles,and the clustering recursive value of calibration result can effectively improve the portability of model parameter calibration result;(2)the method of combining global measurement with clustering recursion to obtain local parameters is the most effective,it can be the optimization research direction of the microscopic traffic simulation model parameter cal

关 键 词：城市交通 仿真参数标定 交叉验证 无信号交叉口 通行能力 

分 类 号：U491[交通运输工程—交通运输规划与管理]