郭宇峥

入学时间：2022级

答辩时间：2025年

论文题目：信号控制交叉口电动自行车交通设计方法研究

中文摘要

摘要

随着电动自行车以其通行机动性强、停车占地少、使用成本低之优势，在城市交通中不断普及并已取代非机动自行车，呈快速增长之势，其高频出行、灵活行驶、行为不确定性等特点与机动车和行人交通矛盾日益突出，特别是在道路交叉口已严重制约了交通通行效率与交通安全。在我国，现状道路交叉口普遍存在电动自行车等非机动车通行空间不足、信号配时关于机动与非机动交通协调考虑不充分等问题，亟需面向电动自行车交通特征系统性地开展城市道路信号控制交叉口交通设计基础与方法研究。为此，本学位论文依托国家自然科学基金项目，以电动自行车交通为研究对象，聚焦信号控制交叉口中通行权配置机制，从“空间—时间—时空协同”三维角度，构建交通设计方法。

本学位论文首先通过实地调研与数据采集，深入解析电动自行车在交叉口的通行时空分布特征和违法违规行为。特别选取电动自行车交通普及率高的南宁市4个典型信号控制交叉口（佛子岭路-翠竹路、云景路-枫林路等），采用无人机航拍（覆盖早高峰7:30-9:00、晚高峰17:00-19:00及平峰时段）与人工观测相结合的方式，采集了交叉口的交通流基础数据，包括机动车与非机动车流量（电动自行车占比79%-91%）、违法违规行为数据、车道宽度和交叉口几何条件，以及信号配时参数等。分析研究了非机动车道宽度、待行区设置、机非隔离和信号配时等要素对电动自行车通行的适应性，以及电动自行车启动快、穿行机动，交通流膨胀与压缩性强等动态特性为信号控制交叉口交通设计方法的以下研究奠定基础：

在空间规划设计层面，提出需求导向的车道资源重构方法。针对非机动车道宽度不足和机非隔离低效问题，建立基于混合整数线性规划（MILP）的车道分配模型，以“通行能力倍增系数”为目标，动态平衡机动车与非机动车道宽度。实例验证显示，在高流量场景（如翠峰路），优化后非机动车道宽度提升，其通行能力倍增系数呈非线性增长，空间资源利用率显著提高。

在时间调控层面，提出了与电动自行车行为适配的信号控制策略和方法。构建了电动自行车“起步延迟—团簇释放—稳定释放”分段通行能力模型（实测拟合R²=0.96），揭示了绿灯初期高释放效率与末期抢行风险基本规律。给出了专用相位（通行信号早启、迟断）与8种错时控制方案，可以有效地实现机非时空分离通行。进一步基于Lane-based模型构建考虑非机动车需求与特征的精细化配时模型，将交叉口整体通行能力倍增系数从0.98提升至1.13，增加约15%，综合提升了交叉口的交通安全与通行效率。

在时空协同层面，提出多场景动态优化方法。构建双层随机优化模型，上层优化车道空间分配以适应多场景需求，下层通过lane-based信号配时实现精细化控制。该模型通过K-means聚类识别典型交通场景，结合样本均值近似法构建多场景输入，实现了车道功能与信号配时的动态联动，在早晚高峰等复杂场景下，整体通行能力倍增系数比只调整车道分配不调整信号配时的策略提升32%，比只调整信号配时，不调整车道分配的策略提升16%。敏感性分析表明，交叉口面积有“黄金宽度区间”，设计时应避免因盲目扩宽导致的时间资源浪费（如50米宽度清空时间增加至7.2秒，有效绿灯时间可大大压缩）。

本研究对以往的交叉口交通设计方法作了创新性探索，特别将电动自行车纳入统一路权分配体系，为有效地提升信号控制交叉口的安全性和高效性提供了理论支撑与实践路径。未来可进一步结合智能交通技术，开展电动自行车交通管理的精细化与智能化问题研究，助力城市交通可持续发展。

关键词：电动自行车，信号控制交叉口，交通设计，时空协同优化，城市交通，lane-based配时

英文摘要

ABSTRACT

With the rapid proliferation of electric bicycles (e-bikes) in urban transportation—owing to their maneuverability, compact parking footprint, and low operational costs—their high-frequency travel patterns, flexible movement, and behavioral unpredictability have increasingly intensified conflicts with motorized vehicles and pedestrians, particularly at signalized intersections, severely compromising traffic efficiency and safety. In China, current intersections face challenges such as insufficient non-motorized traffic space (e.g., e-bike lanes) and inadequate coordination between motorized and non-motorized traffic signal timing. To address these issues, this dissertation, supported by the National Natural Science Foundation of China, systematically investigates traffic design methodologies for signalized intersections by focusing on the right-of-way allocation mechanism for e-bike traffic from three dimensions:spatial planning, temporal control, and spatiotemporal collaboration.

The study first conducts field surveys and data collection at four typical signalized intersections in Nanning City (e.g., Foziling Road-Cuizhu Road, Yunjing Road-Fenglin Road), characterized by high e-bike penetration rates (79%–91% of non-motorized traffic). Drone aerial photography (covering peak and off-peak hours) and manual observations are employed to capture traffic flow data, including vehicle/e-bike volumes, illegal behaviors, lane widths, geometric conditions, and signal timing parameters. Key findings reveal the adaptability of lane width, waiting zones, and signal timing to e-bike dynamics (e.g., rapid acceleration, cluster movement, and traffic flow compressibility), forming the basis for subsequent design methodologies.

Spatially, a demand-driven lane resource reconfiguration method is proposed. A mixed-integer linear programming (MILP) model is developed to dynamically balance motorized and non-motorized lane widths, targeting acapacity multiplier coefficient (μ). Case studies demonstrate that optimized lane widths under high-flow scenarios (e.g., Cuifeng Road) enhance μ nonlinearly, significantly improving spatial resource utilization.

Temporally, signal control strategies aligned with e-bike behaviors are established. A segmented capacity model (“startup delay–cluster release–steady release”) is constructed (R²=0.96), revealing high release efficiency at green onset and end-phase conflict risks. Dedicated signal phases (early green activation/delayed termination) and eight offset control schemes are proposed to achieve spatiotemporal separation. A refinedlane-based signal timing modelfurther elevates the intersection capacity multiplier μ from 0.98 to 1.13 (15% improvement), enhancing safety and efficiency.

Spatiotemporally, a multi-scenario dynamic optimization framework is introduced. A bi-level stochastic model integrates lane allocation (upper level) with lane-based signal timing (lower level). By applying K-means clustering and sample average approximation, the framework dynamically adapts to traffic scenarios, increasing μ by 32% in peak hours compared to lane-only adjustments and by 16% compared to signal-only strategies. Sensitivity analysis identifies a “golden width range” for intersection areas, cautioning against excessive expansion (e.g., 50-meter widths prolong clearance time to 7.2 seconds, compressing effective green time).

This study innovatively integrates e-bikes into unified right-of-way allocation systems, providing theoretical and practical insights for safer and more efficient signalized intersections. Future work should explore intelligent traffic technologies to refine e-bike management, supporting sustainable urban mobility.

Key Words:Electric Bicycles; Signalized Intersections; Traffic Design; Spatiotemporal Collaborative Optimization; Urban Transportation; Lane-based Signal Timing

段行言

入学时间：2022级

答辩时间：2025年

论文题目：城市道路信号控制交叉口非常规车道设置与交通信号协同方法

摘要

中文摘要

随着城市道路交通拥堵加剧和多功能通行需求的不断增加，通过城市道路交叉口设置非常规车道，试图有效利用其时空资源已被认为是提升交叉口效率的重要手段。然而，如何因应需求特征和实际道路交叉口条件最佳地设置非常规车道并同时协同信号配时等尚存在不足，实践应用缺乏有效的理论支撑。为此，本文聚焦于三类典型非常规车道——可变车道、左转外（右）置车道、借道左转车 道，从特征解析，到时空协同优化，再到效果评价的系统性方法，开展以下研究工作：

（1） 非常规车道分类与信号协同系统性框架研究

本文首先基于需求特征从时空资源利用角度，将非常规车道划分为时间复用型与空间结构变型两类，系统地总结其设计理念与适用场景。通过对交叉口交通问题的分类，提炼出适配流向需求不均与交通冲突两类典型要求的非常规车道，并据此构建车道设置与信号控制的时空协同优化框架，明确不同类型非常规车道的设置条件、控制目标与信号配时策略，为后文的车道设置与信号协同优化奠定建模与优化基础。

（2） 考虑流量波动的可变车道设置与控制信号协同方法研究针对高峰期左转与直行流量交替波动的问题，建立了基于延误最小化的可变车道设置判别模型，提出评估交叉口总延误损失与分时段延误损失的精细化方法，并首次引入刻画延误损失的均值与方差双重指标，定量分析流量波动强度与非常规车道设置效益的关联性。在此基础上，考虑不同切换频率与时段颗粒度对可变车道切换策略进行分析，并结合实际数据对未设置可变车道的交叉口给出了分时段设置策略。同时，结合视频数据与实地观测，识别并分析了已设置可变车道切换瞬时的车辆滞留与待转区容量不足等潜在安全问题，评估可变车道实际运行绩效。该部分为交通流波动背景下可变车道设置与信号控制协同优化提供了理论模型与实证支持，增强了可变车道设置的鲁棒性与实用性。

（3） 考虑交通安全冲突的左转外置车道设置与信号协同方法研究针对左转外（右）置车道基础研究不足的问题，系统地梳理了适用的内部与外部冲突场景及应用机制，总结了典型的应用案例。进一步，识别其与常规车道相位结构之间的本质差异，构建涵盖常规车道、左转外置车道及存在外部冲突的交叉口容量解析模型，通过算例对比评估其设置的适用性与效果。进一步提出了基于流量比阈值的设置与否判别方法，揭示左转外置车道适用范围的临界特性。

最后，深入剖析实际失败案例，结合实地调研对左转外置车道安全性特征进行了系统识别与致因分析。

（4） 考虑预信号绿间隔的借道左转车道设置与控制信号协同方法研究针对借道左转车道中左转车辆需提前进入预信号控制区的特性，构建了考虑预信号与主信号绿间隔的通行模型，明确左转预信号灯提前开启与提前关闭的时窗划分。结合信号配时方案，重点探讨了预信号放行时长对借道左转利用效率的影响。基于排队理论与驾驶行为建模，构建了考虑平均交通需求的借道左转车道最佳长度分析模型，并进一步分析了波动交通需求下借道左转车道的利用效率。结果显示利用本模型，交通需求波动时借道左转利用率仍可达60%以上。最后，通过仿真分析与实地观察，识别了借道左转车道潜在安全问题并提出改进对策。

综上，本文系统提出了适用于典型交通需求与问题改善的非常规车道与信号协同控制方法，从时间和空间的双重维度下提升了交叉口的通行效率，丰富并拓展了交叉口交通控制的理论体系，对于实践具有理论意义与实用价值。

关键词：非常规车道；信号协同优化；可变车道；左转外置车道；借道左转车道

英文摘要

ABSTRACT

With increasing urban traffic congestion and the growing demand for multifunctional traffic flows, the implementation of unconventional lanes at urban intersections has been regarded as an effective strategy to enhance intersection efficiency by optimizing the use of time and space resources. However, challenges remain in optimally designing unconventional lanes based on traffic demand characteristics and actual intersection conditions, as well as in coordinating them with signal timing schemes. Existing practical applications often lack robust theoretical support.

To address this gap, this study focuses on three representative types of unconventional lanes—reversible lanes, left-turn external (right-side) lanes, and shared-lane left-turns—and proposes a systematic approach encompassing feature analysis, spatiotemporal coordination optimization, and performance evaluation. The main contributions include:

(1) A Framework for Unconventional Lane Classification and Signal Control

Based on traffic demand characteristics and from the perspective of time-space resource utilization, unconventional lanes are classified into time-sharing types and spatial reconfiguration types. Their design principles and applicable scenarios are systematically summarized.

By categorizing intersection traffic issues, the study identifies two major demand-driven unconventional lane types:those addressing directional flow imbalance, andthose mitigating traffic conflicts.

A spatiotemporal coordination framework for lane design and signal control is developed, clarifying the setting conditions, control objectives, and signal timing strategies for different types of unconventional lanes, thereby laying the theoretical foundation for subsequent modeling and optimization.

(2) Dynamic Lane Design and Signal Control under Flow Fluctuations

Addressing the issue of alternating left-turn and through traffic peaks, a decision-making model based on minimizing total delay is developed. A refined method is proposed to assess both total and time-segmented delay losses, introducing the dual indicators of mean and variance of delay to quantify the relationship between traffic fluctuations and the effectiveness of unconventional lanes.

Based on this, various switching frequencies and time granularities for reversible lane strategies are analyzed. Using real-world data, time-segmented strategies for intersections without dynamic lanes are provided.

Additionally, using video data and field observations, safety issues such as vehicle queuing and insufficient waiting area capacity during lane switching are identified and analyzed, enabling a comprehensive performance evaluation.

This part offers theoretical models and empirical evidence for reversible lane and signal control co-optimization under fluctuating traffic conditions, enhancing robustness and practicality.

(3) Outside Left-Turn Lane Design and Signal Control Considering Conflicts

Given the limited research on outside left-turn lanes, this study systematically analyzes applicable internal and external conflict scenarios and deployment mechanisms, summarizing typical case applications.

The essential differences in phase structures between these lanes and conventional ones are identified. An intersection capacity analytical model is developed to include conventional lanes, outside left-turn lanes, and scenarios with external conflicts. Case studies are used to evaluate applicability and performance.

A threshold-based criterion for implementation, based on traffic flow ratios, is proposed to reveal critical characteristics that define suitable scenarios. Real-world failure cases are dissected through field investigations to systematically identify and analyze safety concerns associated with left-turn external lanes.

(4) Contraflow Left-Turn Lane Design and Signal Control Considering Pre-Signal Green Intervals

To accommodate the need for left-turn vehicles to pre-enter pre-signal control zones in shared-lane left-turn setups, a model is developed that considers green intervals between pre-signals and main signals.

The activation and deactivation windows for the left-turn pre-signal are clearly defined. In conjunction with signal timing strategies, the impact of pre-signal release duration on lane utilization efficiency is explored.

Based on queuing theory and driver behavior modeling, an optimal lane length model is proposed under average traffic demand, followed by an evaluation of efficiency under fluctuating demand. Finally, potential safety risks are identified through simulations and field observations, and corresponding improvement strategies are proposed.

Key Words:Unconventional lanes; Signal optimization; Dynamic lanes; Outside left-turn lanes; Contraflow left-turn lanes; Traffic delay; Safety analysis

何力祺

入学时间：2022级

答辩时间：2025年

论文题目：半响应式模块化自动公交调度优化方法研究

中文摘要

随着城市出行需求日益呈现个性化和多样化特征，传统固定线路公交因缺乏灵活性，已难以满足复杂出行场景下的服务需求。需求响应式公交虽然响应能力强，但由于组织复杂、成本高昂，在高密度区域的应用效果有限。为此，本文聚焦于灵活型公交中的半响应式公交服务模式，即车辆沿着基准线路行驶，并在特定区域内由固定线路模式转化为需求响应式模式，为乘客提供门到门的运输服务。在此背景下，本文结合模块化自动驾驶车辆（Modular Autonomous Vehicles，MAVs）的特性，提出面向MAVs的半响应式公交调度优化方法，旨在提升公共交通系统的响应能力与资源配置效率。本文系统分析了MAVs的物理耦合与解耦机制，使其能灵活响应乘客需求的时空变化，阐明了其与半响应式服务模式的高度契合性。在此基础上，本文引入合乘点布局策略，构建面向偏移区域的两阶段优化框架。第一阶段提出基于改进密度聚类算法的合乘点选址方法，综合考虑了乘客密度、步行距离偏好与固定站点优先；第二阶段建立混合整数规划模型，以最小化车辆行驶时间、乘客车内时间与步行时间为目标，优化MAVs在偏移区域内的行驶路径。

针对运营期内多车次调度问题，提出联合优化发车时刻与MAVs编组数量的调度模型，以最小化系统运营成本和乘客等待时间为目标，构建混合整数线性规划模型，实现调度效率与服务水平的协同提升。

在应用研究方面，本文选取广西钦州市1路公交线路作为应用研究案例，结合真实客流与道路数据，进行调度模拟与效果评估。结果表明，本文提出的半响应式MAVs公交调度方法相较传统固定线路，可有效降低约35%的系统总成本。

此外，本文还通过灵敏度分析评估了发车间隔、模块容量等关键参数对调度结果的影响，验证了模型在不同运营场景下的适应性与稳定性。研究成果可为MAVs系统在实际城市交通体系中的落地应用提供理论支撑与方法参考，具有广泛的推广价值与现实意义。

关键词：公共交通，模块化自动驾驶车辆，合乘点规划，调度优化

英文摘要

ABSTRACT

As urban travel demand becomes increasingly personalised and diversified, traditional fixed-route transit can no longer meet the service demands of complex travel scenarios due to a lack of flexibility. Although demand-responsive transit is highly responsive, its application in high-density areas is limited due to complex organisation and high cost. For this reason, this paper focuses on the semi-responsive transit service model in flexible transit, which means that the vehicle travels along a baseline route and transforms from a fixed-route mode to a demand-responsive mode in a specific area to provide door-to-door transport services for passengers. In this context, this paper combines the characteristics of Modular Autonomous Vehicles (MAVs) and proposes a semi-responsive transit scheduling optimisation method for MAVs, aiming to improve the responsiveness and resource allocation efficiency of the public transport system.

First of all, this paper systematically analyses the characteristics of MAVs to flexibly respond to the spatial and temporal changes of passenger demand through the physical coupling and decoupling mechanism of modular unit vehicles, and clarifies their high degree of compatibility with the semi-responsive service model. On this basis, this paper introduces the strategy of meeting point layout and constructs a two-stage optimisation framework for offset regions. In the first stage, a meeting point selection method based on an improved density clustering algorithm is proposed, which integrates passenger density, walking distance preference and fixed-station preference; in the second stage, a mixed-integer planning model is established to optimise the travel paths of MAVs in the offset region with the objective of minimising vehicle travel time, passenger in-vehicle time and walking time.

Secondly, for the problem of multi-vehicle scheduling during the operation period, a scheduling model is proposed to jointly optimise the departure time and the number of MAVs groups, and a mixed-integer linear planning model is constructed with the objective of minimising the system operation cost and passenger waiting time, so as to achieve a synergistic enhancement of the scheduling efficiency and the service level.

Finally, in terms of application research, this paper selects No.1 bus route in Qinzhou City, Guangxi as an application research case, combining real passenger flow and road data to carry out scheduling simulation and effect evaluation. The results show that the semi-responsive MAVs transit scheduling method proposed in this paper can effectively reduce the total system cost by about 35% compared with the traditional fixed line.

In addition, this paper also evaluates the impact of key parameters such as departure interval and module capacity on the scheduling results through sensitivity analysis, and verifies the adaptability and stability of the model in different operational scenarios. The research results can provide theoretical support and methodological reference for the application of MAVs system in actual urban transport system, which has wide promotion value and practical significance.

Key Words: Public transit，Modular Autonomous Vehicles，Meeting point planning，Scheduling optimization