基于强化协作博弈方法的双车道混合交通流特性

Characteristics of Mixed Traffic Flow in Two-lane Scenario Based on Cooperative Gaming Method

摘要:

对元胞自动机引入Gipps跟驰模型，并结合改进的Q强化学习方法分别建立普通车辆及智能网联车的微观行驶策略，提出了一种新型的混合交通流演化仿真方法.然后，利用数值模拟方式对双车道交通环境进行仿真，探索智能网联车对混合交通流的动态影响.结果表明，相比于元胞自动机构建的普通车辆智能体，改进的Q强化学习方法训练的智能网联车智能体具备更强的连续时空环境适应能力，双车道环境下道路通行能力随着智能网联车渗透率的提升而增大，最高可提升45.34%.此外，智能网联车渗透率的提高会降低车群低效的换道行为，拓宽高通行能力水平下的车辆密度范围，有利于改善交通拥堵.

Abstract:

This paper aims to explore the impacts of connected and automated vehicles (CAV) on traffic flow efficiency based on in-depth microscopic simulation studies using cooperative gaming method. First, the Gipps car-following models were integrated into an improved cellular automata model to mimic the regular vehicle’s driving behavior. Then, cooperative gaming method integrated with enhanced Q-learning was employed as the modeling platform for CAV, to strengthen the capability of the simulation system in realistically reproducing CAV lane changing and car following behavior. Finally, a 2lane freeway stretch was applied to our simulations, and with extensive simulation studies we obtained some promising results. The study results suggest that the impacts of CAV are quite positive. The inclusion of CAV considerably improves traffic flow, mean speed, and traffic capacity. Such understanding is essential for research concerning CAV as well as the CAV implication for future traffic management and control.