基于分位数回归的随机优化速度跟驰模型

为了研究交通流异质性对车辆跟驰行为的影响，基于分位数回归方法改进优化速度函数.根据实际交通流数据对改进的优化速度函数进行参数标定，并对参数结果进行假设检验，结合改进的优化速度函数和全速度差跟驰模型建立随机优化速度跟驰模型，利用傅里叶变换理论推导跟驰模型的稳定性条件，并搭建环形车道仿真平台对跟驰模型进行数值实验.结果表明：改进的优化速度函数能更好地反映交通流异质性对交通流的影响；单一分位点车队达到稳定状态的时间与分位点呈正相关；多分位点组合车队随着0.5分位点车辆数的增加，达到稳定状态的时间减少.提出的多分位点车队相比于单一分位点车队可以更真实地反映交通流复杂的运行状况.

Stochastic optimal velocity car-following model based on quantile regression

The optimized velocity function was improved based on quantile-regression method, in order to study the influence of traffic flow heterogeneity on car-following behavior. The parameters of the improved optimized velocity function were calibrated according to the actual traffic flow data, and the hypothesis test of the parameter results was carried out. Combined with the improved optimized velocity function and the full speed difference car-following model, a stochastic optimized velocity car-following model was established, and the stability conditions of the car-following model were deduced by using Fourier transform theory. A loop lane simulation platform was built to carry out numerical experiments on the car-following model. Results show that the improved optimized velocity function can better reflect the impact of traffic flow heterogeneity on traffic flow. There was a positive correlation between the time when the single quantile team reached the stable state and the quantile. The time to reach steady state was decreased with the increase of the number of vehicles at 0.5 quantile in the multi quantile fleet. Compared with the single quantile fleet, the proposed multi quantile fleet can truly reflect the complex operation of traffic flow.