面向微观交通仿真实验的模型参数敏感性分析方法

为提升微观交通仿真模型标定工作的科学性，对当前微观交通仿真模型标定流程进行梳理，指出微观交通仿真模型标定工作中的敏感性方法应用及定位存在误区。基于微观仿真实验特性及方差分析理论，研究重新构建面向微观交通仿真实验的敏感性分析方法，并界定敏感性分析在仿真模型标定流程中的作用为科学选取校核指标。针对常见信号交叉口进行模型参数敏感性分析，发现选用延误指标作为信号交叉口微观仿真模型标定校核指标才能使得仿真模型标定结果最为有效。

Parameter Sensitivity Analysis Method for Microscopic Traffic Simulation Experiment

The purpose of this research is to improve the science of microscopic traffic simulation model calibration. Based on the combing of existing microscopic traffic simulation model calibration process, this paper suggests there is a certain application error in the function orientation and analysis method of parameter sensitivity analysis which leads to the queried accuracy of calibration result, and the checking index selection was not reasonable. Combining the characteristics of microscopic traffic simulation experiment and the theory of analysis of variance, a new sensitivity analysis method for microscopic traffic simulation specially was put forward and the function of sensitivity analysis in the process of simulation model calibration was repositioned, and its primary role in scientific select checking index was detemined. A common signal intersection was selected and microscopic traffic simulation model parameter sensitivity analysis work was made, the result shows that delay index is appropriate checking index for signal intersection microscopic simulation model calibration.