毕业生知识库

交通运输是国民经济的命脉，直接影响社会经济、 生产与生活的各个方面，与每个人的生活休戚相关。随着城市化进程的不断推进，城市规模不断增大，城市人口越来越多，而分配给交通系统的资源是有限的，交通设施兴建和改善的速度远远赶不上人们日益增长的交通需求，交通拥堵问题日益突出。交通参与者的整体需求与个体需求存在着巨大的差异和明显的矛盾。因此，如何平衡有限的交通资源与不断增长的交通需求之间的矛盾和最大限度地利用好现有的交通资源，成为交通参与者和管理者十分关切的议题。集成了先进信息技术、通信技术、控制技术、传感器技术以及系统综合技术的智能交通系统（Intelligent Transportation Systems, ITS）是平衡交通供需矛盾的有效手段。区域路网交通信号控制是智能交通系统的关键技术，在提高交通资源的利用率，降低能耗，减少交通事故等方面发挥着重大作用。

 

 

本文以解决复杂系统管理与控制的ACP理论为基础，以计算实验方法为手段，结合人工智能领域的自适应动态规划技术，就区域路网交通信号控制问题，分别针对电动车混合车流情况下单交叉口的交通信号控制，干线交叉口的协调控制以及区域网络的协调控制进行了研究。本文的主要工作包含以下几个方面：

 

 

首先，提出了一种基于混合交通流消散模型的单交叉口交通信号控制方法。传统的单交叉口交通信号控制方法多数是基于机动车流而设计的，未考虑电动车对交叉口配时的影响。针对该问题，本文先建立了电动车/机动车混合交通流消散模型，并以此模型为基础，研究了电动车流对交叉口车辆延误的影响，从而推导出混合交通流下的车辆平均延误公式。然后，通过该延误公式得到交叉口最优周期，在此基础上，得到混合交通流下的单交叉口交通信号控制方法。此外，本文还提出了一种混合交通流情况下单交叉口配时的工程应用方法。该方法为电动车比例较高的中小城市的交叉口配时提供了重要参考价值。

 

 

然后，提出了一种基于MILP的半感应干线协调控制方法。针对我国交通流特点，该方法在基于最大化带宽的干线协调方法的基础上，添加了半感应控制策略，能够对交通干线起到良好的控制效果。首先，根据干线交通流特点，利用混合交通流消散模型计算出干线中所有交叉口的周期等参数，获取该干线的关键周期。其次，利用混合整数线性规划得到各交叉口的相序和相位差。然后，在关键周期以及相位差的基础上，添加半感应策略至各交叉口，进一步提高绿灯时间的利用率。最后，通过计算实验对该方法的性能进行评估，计算实验结果表明，该方法能够充分挖掘次相位的绿灯时间，使干线交通流更加平稳，并提高干线车辆的行驶速度，减少车辆的旅行时间。目前，该方法已应用到实际交通系统中。

 

 

最后，提出了基于集成自适应动态规划的区域网络交通信号控制方法。为了应对更复杂的区域网络交叉口的交通信号控制，本文提出了一种集成自适应动态规划（Ensemble Adaptive Dynamic Programming, EADP）方法，该方法无需建立交通系统模型，是一种无模型、自适应的控制方法。首先，本文研究了自适应动态规划（ADP）方法及其在交通控制中的应用情况，分析了其在交通控制中的优点与不足；其次，针对自适应动态规划方法在交通控制中的不足，提出了集成自适应动态规划（EADP）方法，包括一种离线训练策略和三种在线学习策略（Greedy Strategies、Random Boltzman、Weighted Average）。然后，对单交叉口和多交叉口分别设计了相应的集成自适应动态规划控制器。最后，通过计算实验对该方法的性能进行评估，计算实验结果表明，集成自适应动态规划方法可以获得比传统自适应动态规划方法更高的稳定性以及更好的控制效果。

Transportation has an important influence on the social economy and people's quality of life.

Traffic congestion is becoming more and more serious with the increasing urban population and scale. The resources allocated to the transportation system are limited, and the speed of construction and improvement of transport facilities are far behind the increasing traffic demand.

There are huge differences and obvious contradictions between the overall demand and individual demand of traffic participants.

Therefore, how to balance the contradiction between the limited transportation resources and the increasing traffic demand, and to make the best use of the existing traffic resources, has become the subject of great concern to the traffic participants and managers.

The Intelligent Transportation Systems (ITS) is an effective approach to balance the contradictions between traffic supply and demand.

As the key technology of intelligent transportation system, regional control of urban traffic plays an important role in improving the utilization rate of traffic resources, reducing energy consumption and traffic accidents, etc.

 

 

Based on the ACP theory, Computational Experiments method and Adaptive Dynamic Programming techniques, the urban traffic signal control for regional networks is studied in this paper. More specifically, traffic signal control for isolated intersections with e-bicycles mixed traffic flow, arterial coordination traffic signal control, regional coordination traffic signal control are studied respectively. The main work of this dissertation is as follows:

 

Firstly, a traffic signal control method based on the mixed traffic flow discharging model is proposed. Most of the traditional isolated intersection traffic signal control methods are based on the motor vehicle flow, without taking the impact of e-bicycles, which are the main non-motor vehicles in China, into account, when setting intersection signals. In order to solve the problem, an discharging model of e-bicycle mixed traffic flow has been established to explore the vehicle delay effected by e-bicycles. And an average delay formula is developed based on the model.

We analysed the relationship between the average delay and cycle length with the e-bicycle mixed traffic flow.

Additionally, we presented a practical traffic signal control strategy for intersections with e-bicycle mixed traffic flow.

It demonstrates that the optimal cycle length obtained from the proposed method complies with the field application and fits the real traffic circumstances.

 

Secondly, a MILP based semi-actuated arterial coordination method is proposed. Based on the conventional band-maximizing methods, by adding a semi-actuated strategy, this method is a combination of practical actuated traffic control and traditional arterial coordination control. The key cycle length is calculated by mixed traffic flow discharging model, while the phase sequences and offsets are obtained by mixed integer linear programming (MILP). Then the semi-actuated strategy is applied to the scheme in order to improve the utilization rate of green time. Computational experiments are conducted to analyze the performance of the proposed method. The results show that the green time of sub-phase is further utilized, the traffic flow is more smoothly, the mean speed and travel time are improved. This method has been applied into a practical traffic system and field test proves the effectivity of the method.

 

Thirdly, an Ensemble Adaptive Dynamic Programming (EADP) method is proposed to solve the regional traffic signal control problem. The method can

improve the generalization ability and stability of traditional ADP controller by training a finite number of sub-ADP controllers and combining their results. It is helpful for traffic scientists and engineers to solve real-world traffic control problems. As a kind of intelligent control methods, EADP has great potential for other complex nonlinear systems in solving the optimal control problem. Computational experiments show that  the EADP traffic signal controller is better

than a single ADP controller and the traditional fix-time controller.