首页|基于多国实测数据的跟驰模型对比

基于多国实测数据的跟驰模型对比

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
为了更精确描述车辆跟驰(CF)行为,并研究不同国家跟驰行为数据对跟驰标定模拟的影响,以及各跟驰模型对跟驰行为模拟的精确程度,选取中国西安市南二环某路段交通流CHD数据集、美国NGSIM数据集以及德国HighD数据集,针对Gazis-Herman-Rothery(GHR)模型、智能驾驶模型(IDM)以及最新被提出的S-shaped three-parameters(S3)跟驰模型进行模型标定以及误差分析,利用加速度、前后车速度差、前后车位置差和后车速度等数据作为输入参数,采用互相关分析与模拟退火相结合的方法进行数据拟合,并利用加速度、速度和位移的均方根误差(RMSE)对参数拟合后的模型进行性能评价.研究结果表明:针对3个不同国家数据集中的跟驰行为,S3微观模型标定效果均表现最佳,3个数据集的RMSE平均值均最小,且低于其他2种跟驰模型;德国HighD数据集总采集精度高、数据量大,因此无论采用何种CF模型进行标定,该数据集在跟驰行为标定方面的性能均表现最佳、误差最小.研究结果对交通仿真软件模拟交通流的车辆跟驰模型选取及其参数优化具有重要意义,且对于如何选择跟驰模型标定数据集亦具有重要价值.
Contrastive of car-following model based on multinational empirical data
In order to describe the car-following(CF)behavior of vehicles more accurately,and to study the influence of car-following behavior data from different countries on the calibration of car-following models,as well as the accuracy of various car-following models in simulating car-following behavior,three datasets was selected as a traffic flow dataset from CHD dataset from a section of the South Second Ring Road in Xi'an,China,the U.S.NGSIM dataset,and the German HighD dataset.The Gazis-Herman-Rothery(GHR)model,the intelligent driver model(IDM),and the newly proposed S-shaped three-parameters car-following model(S3)were used for model calibration and error analysis.Acceleration,speed difference between the leading and following vehicles,position difference between the leading and following vehicles,and the speed of the following vehicle were used as input parameters.A combination of cross-correlation analysis and simulated annealing methods were employed for data fitting.The performance of the fitted models was evaluated using the root mean square error(RMSE)of acceleration,speed and displacement.The results show that for the car-following behavior in the three different countries'datasets,the S3 microscopic model shows the best calibration performance,with the lowest average RMSE for all three datasets compared to the other two car-following models.Due to the high overall data collection accuracy and large data volume of the German HighD dataset,it exhibits the best performance and lowest error in car-following behavior calibration,regardless of the car-following model used.The research results are of great significance for the selection of car-following models and parameter optimization in traffic simulation software,and hold important value for the choice of datasets for car-following model calibration.4 tabs,10 figs,30 refs.

traffic engineeringmicroscopic traffic flowcar following modelS3 modelHighD datasetsimulated annealing method

徐志刚、魏璐颖、刘志广、刘张琦、秦孔建

展开 >

长安大学信息工程学院,陕西西安 710064

陕西汽车控股集团有限公司,陕西西安 710299

中汽科技(北京)有限公司,北京 100176

交通工程 微观交通流 跟驰模型 S3模型 HighD数据集 模拟退火法

国家重点研发计划国家自然科学基金陕西省自然科学杰出青年基金陕西省自然科学基础研究计划青年项目

2019YFB1600101619730452023-JC-JQ-452023-JC-QN-0667

2024

10.19721/j.cnki.1671-8879.2024.02.009

长安大学学报(自然科学版)
长安大学

长安大学学报(自然科学版)

CSTPCD北大核心
影响因子：1.011
ISSN：1671-8879
年,卷(期)：2024.44(2)
  • 30