快速的城市化与机动化造成了严峻的交通拥堵问题。公共交通被认为是缓解城市交通拥堵的有效方法。在常规公交中，公交信号优先是实施公交优先发展的重要举措之一。受制于有限的信息获取渠道以及单一的控制手段，传统公交信号优先控制方法在许多场景下实施效果不佳，而车路协同系统可以为摆脱上述困境提供一个有效的途径。基于先进的无线通信技术和丰富的数据采集手段，车路协同环境可以提供更加多源、准确、实时的交通数据，从而可以实施有效的车辆速度引导和信号控制，进而能够科学地减缓公交车和非公交车辆间的矛盾。本文的主要对如下内容展开了研究：

(1) 车路协同环境下公交信号优先的实施框架。难以稳定为公交车提供优先通行权、难以有效控制非公交车辆所受的负面影响以及难以有效降低公交车的全程信控延误是现有公交信号优先控制方法的主要不足。为了在车路协同环境下实施更加稳定而有效的公交信号优先控制，车路协同系统应当具备由中央控制单元、交叉口控制单元、公交车到达检测单元、停靠时间统计单元、交通流流量检测单元、交通波波速估计单元和信号控制单元组成的智能路侧子系统、由定位与授时单元和速度控制单元组成的智能车载子系统和由路侧通信单元和车载通信单元组成的无线通信子系统。车路协同环境提供的公交车的历史驻站停靠时间、实时交通波波速和各车道实时车辆到达等信息以及速度引导等控制手段被用于完善现有公交信号优先控制方法的不足。

(2) 有公交专用道的单个交叉口公交信号优先控制。公交车驻站停靠时间的随机性导致公交车抵达交叉口停止线的时刻难以准确预测，而基于不精准的预测结果设计的公交信号优先控制方法失效概率较高。本文构建了一种基于周期伸缩策略，考虑公交车驻站停靠时间随机性的公交信号优先控制方法，此方法以公交车的期望信控延误和非公交车辆的平均信控延误为优化目标。周期伸缩策略通过同时调整信号周期长度和各个相位的绿信比实现更灵活的绿灯时长再分配，以达到降低拒绝公交信号优先请求概率的目的。利用车路协同环境提供的实时车辆到达信息，非公交车辆的平均信控延误可通过有限状态自动机计算；利用车路协同环境提供的公交车驻站停靠时间的概率分布，公交车的期望信控延误可通过时空图计算。NSGA Ⅱ(Nondominated Sorting Genetic Algorithm 2)算法被用于求解帕累托前沿，基于帕累托前沿而提出的决策规则可实现公交车优先等级的动态控制。VISSIM仿真实验结果表明，与传统公交信号优先控制方法(Conventional Transit Signal Priority, CTSP)几乎失效相比，在高、中、低三种交通拥堵水平下，此方法可将公交车平均信控延误降低27%~47%，且将非公交车辆的平均信控延误降低了约13%，较充分地发挥了公交专用道的作用。此外，在单次优先中，不同优先等级下，公交车的平均信控延误的极差和相对极差可达30秒和160%，而非公交车辆的平均信控延误的极差和相对极差可达10秒和15%，表明此方法在优先等级控制上的效果。

(3) 无公交专用道条件下的单个交叉口公交信号优先控制。无公交专用道条件下，由于受到车辆排队的影响，公交车抵达交叉口停止线的时刻不能基于匀速运动假设计算。本文基于周期伸缩策略构建了一种构建面向无公交专用道条件的公交信号优先控制方法，此方法以公交车期望信控延误和信号调整幅度的加权和为优化目标。为计算公交车信控延误，本文将无公交专用道条件下公交车的行驶过程划分为“排队等候”、“停站上下客”、“驻站淹没”、“抵达停止线”和“切换周期”五个事件。每个事件内公交车产生的信控延误可基于车路协同环境提供的实时交通波波速计算。各事件的发生顺序则由事件驱动法动态确定，所有事件产生的信控延误之和即为公交车信控延误。COBYLA (Constrained Optimization By Linear Approximation)算法被用于求解此非凸信号优先优化模型。VISSIM仿真实验结果表明，与CTSP几乎失效相比，在高、中、低三种交通拥堵水平下，此方法均能以约5%的非公交车平均信控延误提升为代价，将公交车平均信控延误降低63%以上，适用于无公交专用道条件下的政策性优先场景。

(4) 干线协同式公交信号优先控制。无条件优先通常会造成非公交车辆信控延误显著提升，而条件优先不能确保公交车优先通过所有交叉口，各交叉口独立实施公交信号优先时，公交车可能因在部分交叉口不满足优先条件而停车等候，进而导致公交车全程信控延误未显著降低。本文构建了一种有公交专用道的条件下的干线协同式公交信号优先控制方法，此方法通过广义相位插入策略和车路协同环境提供的速度引导机制干预公交车通过当前和下游交叉口的时刻。通过极小化绿灯错位度此方法可使公交车尽可能仅通过速度引导实现不停车通过下游交叉口。ECP (Extended Cutting Plane)算法被用于求解此非线性混合整数信号优先优化模型。VISSIM仿真实验结果表明，与CTSP仅将公交车全程信控延误降低26%左右相比，在各交通拥堵水平下，此方法均能以不足8%的非公交车平均信控延误提升为代价，将公交车平均信控延误降低85%以上，几乎完全消除公交车的信控延误。由于未考虑公交车站的影响，本方法适用于BRT(Bus Rapid Transit)等拥有独立路权且公交车站站距较大的公交线路。

关键词：车路协同系统，公交信号优先，公交车站，公交专用道

Rapid urbanization and motorization have resulted in severe traffic congestion. Public transport is regarded as an effective way to relieve traffic congestion in cities. In the conventional bus, transit signal priority (TSP) is one of the important measures to implement bus priority development. Due to the limited information acquisition channels and single control means, the traditional transit signal priority control method is not effective in many scenarios, and the Intelligent Vehicle Infrastructure Cooperative Systems (IVICS) can provide an effective way to get rid of the above predicament. Based on advanced wireless communication technology and abundant data acquisition means, IVICS can provide more multi-source, accurate and real-time traffic data, so as to implement effective vehicle speed guidance and signal control, and thus scientifically mitigate the contradiction between transits and non-transits vehicles. The main contents of this paper are as follows:

(1) Implementation framework of transit signal priority control in the IVICS environment. It is difficult to provide priority for buses stably, difficult to effectively control the negative impact of non-bus vehicles and difficult to effectively reduce the total signal delay of buses are the main shortcomings of the existing transit signal priority control method. In order to implement more stable and effective transit signal priority control, the Intelligent Vehicle Infrastructure Cooperative Systems shall have intelligent roadside subsystem composed of central control unit, intersection control unit, bus arrival detection unit, stop time statistics unit, traffic flow detection unit, traffic wave speed estimation unit and signal control unit, intelligent vehicle-mounted subsystem composed of positioning and timing unit and speed control unit, and  wireless communication subsystem composed of roadside communication unit and vehicle-mounted communication unit. The bus historical stop time, real-time traffic wave speed, the number of vehicles arriving in each lane and other information provided by the IVICS environment as well as control means such as speed guidance are used to improve the shortcomings of the existing transit signal priority control methods.

(2) Transit signal priority control at single intersection with exclusive bus lanes. The stochasticity of bus dwell time makes it difficult to accurately predict the moment when buses arrive at the intersection stop line, and the transit signal priority control method designed based on the inaccurate prediction has a high probability of failure. Considering the stochasticity of bus dwell time, a transit signal priority control method based on the variable length strategy is constructed in this paper. The signal priority optimization model of this method takes the expected signal delay of buses and the average signal delay of non-bus vehicles as the optimization objectives. By adjusting the cycle length and the green signal ratio of each phase at the same time, the variable cycle length strategy can realize more flexible green time reallocation, so as to reduce the probability of rejecting TSP requests. The average signal delay of non-bus vehicles is calculated based on finite state machine using real-time vehicle arrival information provided by the IVICS environment, and the expected signal delay of buses is calculated based on the probability distribution of bus dwell time provided by IVICS based on space-time graph. NSGA Ⅱ (Nondominated Sorting Genetic Algorithm 2) algorithm is used to solve the Pareto frontier. The proposed decision rules based on the Pareto frontier can dynamically control the level of bus priority. The results of VISSIM simulation show that, compared with Conventional Transit Signal Priority (CTSP), this method can reduce the average bus signal delay by 27%~47% under high, middle and low traffic congestion levels. Moreover, the average delay of non-bus vehicles is reduced by about 13%, which means the method effectively plays the role of the exclusive bus lanes. In addition, in a single priority, the average range and relative range of bus delay can reach 30 seconds and 160% among different priority levels, while the average range and relative range of non-bus delay can reach 10 seconds and 15%, indicating the effectiveness of this method in priority level control.

(3) Transit signal priority control at single intersection without exclusive bus lanes. Under the condition of no exclusive bus lane, due to the influence of vehicle queuing, the moment when the bus arrives at the stopping line at the intersection cannot be calculated based on the assumption of uniform motion. In this paper, based on the variable cycle length strategy, a transit signal priority control method is constructed for the condition without exclusive bus lanes. The signal priority optimization model of this method takes the weighted sum of bus expected signal delay and signal adjustment amplitude as the optimization objective. In order to calculate the bus delay, this paper divides the bus running process under the condition without exclusive bus lanes into five events: "waiting in line", "getting on and off passengers at the stop", "flooded at the stop", "arriving the stop line" and "switching cycle". The signal delay generated by buses in each event can be calculated based on the real-time traffic wave speed provided by the IVICS environment. The occurrence sequence of each event is determined dynamically by the event-driven method, and the sum of delays generated by all events is the bus delay. COBYLA (Constrained Optimization By Linear Approximation) algorithm is used to solve the non-convex optimization model. The results of VISSIM simulation experiments show that the CTSP almost failed, while this method can reduce the average bus delay by more than 63% at the cost of about 5% increase in the average non-bus delay under high, middle and low traffic congestion levels. It is suitable for the policy-oriented priority scenarios without exclusive bus lanes.

(4) Coordinated transit signal priority control for arteries. Unconditional priority usually results in significant increases in signal delays for non-bus vehicles. However, conditional priority cannot ensure that buses pass all intersections preferentially. When transit signal priority is independently implemented at each intersection, buses may stop and wait because they do not meet the priority requirements at some intersections, thus leading to no significant reduction in bus signal delay throughout the whole trip. In this paper, a coordinated transit signal priority control method for arteries with exclusive bus lanes is constructed. In this method, the generalized phase insertion strategy and the bus speed guidance mechanism provided by the IVICS environment are used to intervene the moment when the bus passes the current and downstream intersections. The signal priority optimization model takes the weighted sum of signal adjustment amplitude, bus travel time and green dislocation degree as the optimization objective. By minimizing the degree of green dislocation, this method can make buses pass downstream intersections without stopping through speed guidance as far as possible. The Extended Cutting Plane (ECP) algorithm is used to solve this nonlinear mixed-integer optimization model. The results of VISSIM simulation experiments show that, compared with CTSP, which only reduces the whole bus signal-control delay by about 26%, this method can reduce the bus signal delay by more than 85% at the cost of less than 8% of the non-bus signal delay under all traffic congestion levels, and almost completely eliminate the bus signal delay. Since the influence of bus stops is not considered, this method is suitable for Bus Rapid Transit (BRT) and other bus lines with independent road rights and large distance between bus stops.

Key Words: Intelligent Vehicle Infrastructure Cooperative Systems (IVICS), Transit Signal Priority (TSP), Bus stop, Exclusive bus lane