【作者】 蒋海峰；

【导师】 魏学业；

【作者基本信息】 北京交通大学 ， 交通信息工程及控制， 2008， 博士

【摘要】 混合交通是指机动车、自行车和行人混杂在一起的交通，它是我国大多数城市平面信号交叉口典型的交通特征。机动车和非机动车在平面信号交叉口处相互之间的冲突和干扰现象，严重影响了交叉口的通行效率，增加了交通安全隐患。而混合交通是一个非线性、强随机性和时变性的复杂系统，要解决城市混合交通问题，需要对城市道路平面交叉口处的混合交通流进行系统研究。本研究的目的是针对平面信号交叉口混合交通中的机非干扰机制、微观建模方法和非线性特性等关键问题进行深入研究，建立混合交通流的若干微观行为模型，为提高交叉口的通行效率和交通管理提供理论依据。首先，本文对信号交叉口机非混合交通流的干扰机理进行了深入细致的研究，分析了信号交叉口内机非混合交通流的干扰机理及其特性，提出了干扰区和干扰程度的概念，为机非干扰问题的定性和定量分析提供了理论依据。基于非线性跟驰理论建立了右转机动车进入交叉口前的减速模型、在交叉口内的跟随行驶模型和加速驶离模型。通过研究机动车在信号交叉口处的微观行为特性，建立了右转机动车穿越自行车区的行为决策模型。通过分析自行车在交叉口内的纵向前进和横向移动行为特征，建立了单辆自行车在交叉口的减速、停车、加速、驶离和超车等模型，以及自行车群二维微观行为仿真模型。研究了交通流混沌现象产生机理及其动态变化规律，基于混沌时间序列分析理论，提出了一种快速计算短时交通流时间序列最大Lyapunov指数的小数据量方法，并结合庞卡莱截面法对识别结果进行了验证。利用最小覆盖维方法分析了信号交叉口短时交通流的局部分形特性，确定了最小覆盖维数与短时交通流量波动的关系，提出了一种能估计短时交通流量发展和变化趋势的最小覆盖维方法。基于Multi-Agent理论，提出了信号交叉口混合交通流微观仿真系统框架，通过计算机仿真再现了信号交叉口混合交通流的微观动态行为特性，实测数据和仿真分析结果验证了所建模型是正确的。更多还原

【Abstract】 Mixed traffic, composed with motor vehicles, bicycles and persons, is a typical characteristic in China city. The interference and conflict phenomenon, between motor vehicles and bicycles, reduce the traffic capacity of intersections severely and increase the hidden traffic troubles. Mixed traffic is a complicated system with serious nonlinear, strong stochastic and time varying characteristics. To solve the problem of urban mixed traffic, it is needed to systemic study on mixed traffic flow. This research is aims to study the mixed traffic flow and interference principles at signalized intersections, micro behavior models and nonlinear dynamic characteristics and other key problems. Some micro behavior models of mixed traffic flow are developed. This can provide basis to improving the traffic efficiency and management of signalized intersections.First, the interference principles of mixed traffic flow at signalized intersections have been studied and the characteristic of interference principles are also analyzed. The conceptions of interference region and interference grade are proposed, which will provide theoretical basis for qualitative and quantitative analysis of interference questions. Based on nonlinear car-following theory, the deceleration model before intersections, car-following model at intersections and the acceleration model departure from intersections of right-turning motor vehicles are developed. Through researching the micro behavior characteristics of motor vehicles at signalized intersections, a behavior judgment model about right-turning motor vehicles traversing straight-going bicycles is proposed and developed.The single bicycle simulation models, including deceleration and parking, acceleration and departure as well as traversing, are established by researching the characteristics of transverse and lateral movement at intersection. And a two-dimensional simulation model is also established to represent bicycle group’s micro behavior at signalized intersections.The generation principles of traffic flow’s chaotic phenomena and its dynamic characteristic have been studied. A little data volume method is proposed to fast calculating shot period traffic flow time series maximum Lyapunov exponent based on chaotic time series. The identification outcome is verified by Poincare section. The part fractal characteristic of short period traffic flow is analyzed using minimal cover dimension. The relationship between minimal cover dimension and short period traffic flow fluctuation is determined. And a minimal cover dimension method that can estimate short period traffic flow variety is proposed.A micro simulation system framework of mixed traffic at signalized intersections is developed based on multi-agent theory. The micro behavior characteristics of mixed traffic flow are re-showed by computer simulation. These models are validated separately with field data and simulation results.更多还原