交织区微观交通仿真模型参数敏感性分析  

Parameter Sensitivity Analysis of Microscopic Traffic Simulation Model in Weaving Area

作　　者：崔欣宇 宋霞飞 周晨静[1] CUI Xin-yu;SONG Xia-fei;ZHOU Chen-jing(General Aviation Beijing Laboratory,Beijing University of Civil Engineering and Architecture,Beijing 102616,China)

机构地区：[1]北京建筑大学通用航空北京实验室,北京102616

出　　处：《科学技术与工程》2023年第27期11854-11861,共8页Science Technology and Engineering

基　　金：国家自然科学基金青年科学基金(51708017);北京建筑大学研究生创新项目(PG2022033)。

摘　　要：开展微观仿真模型参数标定是科学应用微观仿真技术的前提,也是开展快速路交织区仿真模拟,继而挖掘交通流运行规律的必要步骤。参数敏感性分析是筛选标定参数种类的核心步骤。研究开展了针对交织区不同流量状态下不同指标的参数敏感性问题。根据基本道路通行能力设定交织区仿真中低、中、高三种流量条件,在不同的流量条件下开展交织区的参数敏感性分析工作。同时考虑交织区交通运行特点,分别从速度、流量和延误几个层面选取六个校核指标作为微观驾驶行为参数敏感性的分析对象。以VISSIM为平台设计并开展微观仿真实验,实践全样本空间下机器学习方法开展参数敏感性分析的有效性。对比不同参数敏感性分析方法并综合实验结果从车辆运行特征角度解释不同流量状态、不同指标的参数敏感性结果,证明以延误为校核指标、以交通流近饱和状态为标定场景,可以获得更全面的标定结果。The calibration of microsimulation model parameters is a prerequisite for scientific application of microsimulation technology,and also the necessary step for simulating weaving areas of urban expressway to mine the traffic flow operation rules.Parameter sensitivity analysis is the core step for screening the types of calibration parameters.The study was carried out to investigate the sensitivity of parameters for different indicators in the weaving area under different traffic conditions.According to the basic road capacity,three flow conditions of low,medium and high were set in the simulation of weaving area to carry out parameter sensitivity analysis of the weaving area.At the same time,considering the traffic operation characteristics of the weaving area,six check indicators were selected from speed,flow and delay levels as the objects of microscopic driving behavior parameter sensitivity analysis.Micro simulation experiment with VISSIM as the platform was designed and carried out.The validity of machine learning method was verified in the full sample space.The parameter sensitivity results of different flow states and different indexes are explained from the perspective of vehicle running characteristics,which is under comparing different parameter sensitivity analysis methods and combining experimental results.It is proved that more comprehensive calibration results can be obtained by taking delay as the checking index and traffic flow near saturation state as the calibration scene.

关 键 词：交通工程 交织区 微观交通仿真 参数敏感性分析 机器学习 VISSIM 

分 类 号：U491[交通运输工程—交通运输规划与管理]