无人驾驶车辆对道路通行能力的影响机理及作用研究

Traffic capacity is an important basis for road traffic planning, design and management. Driving behavior characteristics of driverless vehicles are quite different from those of ordinary vehicles, which will definitely bring changes to traffic flow and road capacity after merging into ordinary traffic flow. Based on the research results of predecessors on ADAS, ACC, CACC and traffic flow microscopic models, this project analyses the sensing technology, driving scene identification and operation logic of driverless vehicles at home and abroad. Furthermore, the driving behavior characteristics of driverless vehicles and ordinary vehicles under the condition of man-machine mixed driving are analyzed. Combined with simulation analysis of heterogeneous traffic flow stability, the following models of SV-SV, SV-HV, HV-SV and HV-HV are optimized and improved. Considering the characteristics of cooperative control of driverless vehicles, clustering effect parameters are introduced to build lane change models of driverless vehicles and ordinary vehicles. Based on VISSIM secondary development technology, a man-machine mixed driving traffic flow model is established. The road test data of driverless vehicles and the measured data of ordinary vehicles are used as correction data, and the SPSA algorithm is used to realize the automatic correction of model parameters. Based on the import rate of driverless vehicles and clustering effect, the traffic flow microscopic simulation method is used to quantitatively analyze the road capacity and reveal the influence mechanism and function of man-machine driving interaction on the road capacity.

道路通行能力是道路交通规划、设计、管理的重要依据，无人驾驶车辆与普通车辆的驾驶行为特性截然不同，汇入普通车流后必将给交通流及道路通行能力带来变化。本项目通过借鉴前人对ADAS、ACC、CACC等技术及交通流微观模型的研究成果，分析国内外无人驾驶车辆的传感技术、驾驶情景辨识与操作逻辑，分析人机混驾条件下无人驾驶车辆、普通车辆的驾驶行为特征；结合异质交通流稳定性仿真分析，优化改进SV-SV、SV-HV、HV-SV、HV-HV跟驰模型；考虑无人驾驶车辆协同控制的特点，引入聚类效应机制，构建无人驾驶车辆和普通车辆的变道模型；基于VISSIM二次开发技术，建立人机混驾交通流模型，采用无人驾驶车辆路测数据及普通车辆实测数据做校正数据，运用SPSA算法实现模型参数的自动校正；采用交通流微观仿真法，基于无人驾驶车辆汇入率及聚类效应进行道路通行能力定量分析，揭示人机驾驶相互作用对道路通行能力的影响及作用。

本项目主要针对高速公路人机混驾交通流中无人驾驶车辆和普通车辆的跟驰、变道模型构建、人机混驾交通流模型构建、及道路通行能力仿真分析展开研究，包括：（1）人机混驾条件下驾驶行为特性分析；（2）人机混驾条件下跟驰模型构建；（3）人机混驾条件下变道模型构建；（4）人机混驾交通流模型构建及参数标定；（5）无人驾驶车辆对道路通行能力的影响机理及作用。.基于计划行为理论，利用AMOS24.0软件分析,实现了驾驶员在跟驰网联自动驾驶车辆时跟车间距变化意向的量化；结合问卷调查，实现了驾驶员在跟驰网联自动驾驶队列时跟车间距的量化；分析了人机混驾条件下无人驾驶车辆、普通车辆的驾驶行为特征，结合异质交通流稳定性仿真分析，优化改进了SV-SV、SV-HV、HV-SV、HV-HV跟驰模型；考虑无人驾驶车辆协同控制的特点，引入聚类效应机制，完成了无人驾驶车辆和普通车辆的变道模型构建；研究了传统车辆、ACC车辆和CACC车辆混行所构成的复杂异质交通流稳定性，针对复杂异质交通流，应用了基本图模型，推导出基于临界速度的通行能力模型，分析在不同CACC车辆比例下的通行能力。分析反应时间、期望车间时距和传统车辆比例对通行能力的影响，提出了通行能力提升策略。.相应成果在《深圳大学（理工版）》、《Physica A: Statistical Mechanics and its Applications》、《交通信息与安全》、《交通运输系统工程与信息》、《安全与环境学报》、《中国公路学报》、《Journal of Advanced Transportation》、《TRANSPORTATION RESEARCH BOARD》等学术期刊以及国际会议上发表。发表学术论文7篇，其中SCI检索2篇次， EI检索2篇次；申请国家发明专利6项，其中授权2项；获准计算机软件著作权2项；培养硕士研究生4人毕业获工学硕士学位，4人在读。