微观仿真参数敏感性自动化分析方法与实证

Automatic Analysis Method and Demonstration of Micro Simulation Parameter Sensitivity

微观仿真模型参数标定是仿真技术得以科学应用的前提,针对微观仿真模型参数敏感性分析过程繁杂、实验量大、消耗时间长等问题,提出一种基于决策树方法的微观仿真参数敏感性分析的自动化实现方法,包含实验设计、仿真模型赋值与运行、结果汇总和敏感性分析四大模块。研究对各个模块的实现流程进行详细说明,并以快速路交织区为仿真场景,实证分析微观仿真模型中各类参数对宏观交通流延误、车辆运行速度以及交通流率这三个指标的敏感性。

Micro simulation model parameter calibration is a prerequisite for the scientific application of simulation technology.To address the problems of complicated process,large amount of experiments and long time consumption of micro simulation model parameter sensitivity analysis,an automated implementation method of micro simulation parameter sensitivity analysis based on decision tree method is proposed,including four modules of experimental design,simulation model assignment and operation,result summary and sensitivity analysis.The study provides a detailed description of the implementation process of each module and empirically analyzes the sensitivity of various parameters in the micro simulation model to the three indicators of macroscopic traffic flow delay,vehicle operating speed and traffic flow rate using the expressway intertwining area as the simulation scenario.