基于低频定点检测数据的交叉口交通状态估计

Traffic State Estimation based on Low Frequency Detection Data at Signalized Intersections

摘要:

本文针对我国中小城市数据现状, 提出了一种基于路中定点线圈低频(60s频率)检测数据的交叉口交通状态估计方法。该方法基于仿真数据，分析了不同环境变量组合条件下占有率、流量和交通状态的关系，并提出了基于线性拟合的交通状态分界线建立方法；又利用多元线性回归拟合出分界曲线各系数与环境变量的函数关系，用其估计一般条件下的交通状态。经过验证，本方法仿真环境和实证环境下的平均估计准确率分别达到80%和75%以上，且严重错误率均低于2.1%。

Abstract:

A traffic state identification method for intersections is proposed based on detection data with a low frequency from detection on the mid of the urban roads which is applied for urban interrupted flow in medium and small cities of our country. At first, the relationship between occupation, volume and traffic state is analyzed under different parameters of circumstances based on simulation data and a method of curve fitting is proposed to build the boundaries of different traffic state. Then the functional relationship of coefficients of boundaries functions with environment variables is fit out which is later applied to general ones. The methods above is verified by simulation data with the identification rate of over 80%, and empirical data with the identification rate of over 75% , with severe mistake rates less than 2.1%.