车联网环境下匝道汇入区瓶颈换道优化

Freeway Merging Area Lane Changing Advisory Optimization Under Connected Vehicles Environment

摘要:

高快速路汇入区（即合流区）瓶颈是交通流运行的咽喉，汇入瓶颈交通流失效会加剧拥堵，诱发交通振荡以及事故率上升等一系列问题。与现有研究大都通过调节匝道汇入车辆行为或主线车辆速度进而试图改善汇入区交通流问题不同，该研究聚焦于瓶颈汇入区上游主线车辆，通过动态调节汇入区上游主线车道车辆分布，提升汇入区通行能力。具体而言，研究提出一种可以对网联车（CV）进行双向换道建议的混合整数线性规划模型，该方法不依赖于交通流基本图设定的临界密度，通过实时计算每一辆个体CV的向左、向右或保持车道决策以优化车道流量分布，减少汇入车辆干扰，提升汇入效率。基于VISSIM交通仿真软件，通过二次开发搭建了汇入区瓶颈换道优化实时仿真评估系统，并对该方法进行了验证，测试不同流量组合和不同CV渗透率下算法的有效性。各车道时空轨迹表明该换道建议优化方法可以有效减小汇入车辆冲突，车均延误分析结果表明在单车道平均流量1 550 ~1 800 veh·h^-1区间，即汇入瓶颈失效关键流量区段，换道建议优化方案相比原方案能显著改善汇入区的运行效率，车均延误可降低10 %~50 %左右。CV渗透率敏感性分析表明，在较低的0.2~0.5渗透率下即可达到减小延误的目标。

Abstract:

The on-ramp merging area is the bottleneck of the freeway system， the breakdown of which will trigger many problems such as the deterioration of traffic congestion， traffic oscillation， and increase in accident rate. Currently most studies aim to improve the operation of merging area by adjusting the merging behavior of on-ramp vehicles or controlling the speed of the mainline vehicles. This research， on the other hand， focus more on the potential of mainline vehicles in lifting the traffic throughput by dynamically controlling the upstream lane flow distribution. Specifically， a lane changing advisory optimization model under the connected vehicles （CV） environment is proposed， in which the lane changing decision （change to the left/right lane or keep the current lane） of each individual CV is optimized to reduce the disruption caused by the merging vehicle and thus improve the efficiency. The optimization model is formulated as a mixed integer linear programming，which is independent of the critical density derived from fundamental diagram. Based on traffic simulation software VISSIM， a real-time evaluation platform is developed， in which the proposed lane changing advisory method is implemented and further validated under various traffic demand combinations and CV penetration rate. Trajectories from all the lanes show that this method could effectively reduce the conflict of merging vehicles. The averaged traffic delay analysis results reveal that the proposed method can significantly improve the operation efficiency if the average traffic flow level is near the critical flow of traffic breakdown， i.e.， 1 550 –1 800 veh·h^-1 per lane. In such scenarios， the average traffic delay could be decreased by 10 %–50 %. Furthermore， the sensitivity analysis on CV penetration rate show that at a with lower CV penetration rate of 0.2–0.5， the proposed method is capable of reducing traffic delay.