信号交叉口行人过街风险评估模型研究

The "Chinese way to cross a street" has been heated discussed recently, it is urgent to pay attention to the high frequency and severity of pedestrian accidents happened at signalized intersections. A paradox that "it is even more dangerous to cross on Green than on Red" reflects deficiencies of traffic facility settings. Therefore, it is critical to recognize pedestrian risk sources and causes accurately. Under the conditions of lacking sufficient accident data in China, the applicants propose a method to recognize pedestrian safety factors based on risk analysis, and develop pedestrian risk assessment models from both macroscopic and microscopic levels. In the macroscopic level, the applicants will do a life-cycle analysis of intersection planning, design, operation and control, in order to find out influencing factors on pedestrian safety and assess them quantitatively. Fuzzy DEMATEL method will be used to select the "source" factors. The macroscopic model helps to assess pedestrian overall safety at intersections rapidly. In the microscopic level, split the cycle length and crosswalk into several time and space intervals, calculate pedestrian-vehicle (bicycle) conflict risk degree by using event tree and Monte-Carlo method, analyze its time-space distribution, and finally assess pedestrian risk dynamically based on the whole process of pedestrian crossing. The microscopic model can be used to assess pedestrian safety of whom with different behavior characteristics. The research results are expected to provide theoretical support for a proper intersection planning, design, operation and control, which can eliminate or reduce pedestrian risk and achieve proactive risk prevention.

"中国式过马路"成为近年来热议的话题，信号交叉口行人事故的高发性和严重性不容忽视。"行人绿灯过街反比红灯过街更危险"的怪现象折射出交通设施在设置上存在明显缺陷，准确辨识信号交叉口行人过街风险的来源和成因是解决该问题的关键。本项目拟在国内全样本交通事故数据匮乏的条件下，建立一套基于风险分析的行人安全因素识别方法，并开发宏微观相结合的行人过街风险评估模型。宏观上，建立交叉口规划设计、运行管理全生命周期的行人风险因素主动识别与风险量化方法，采用模糊DEMATEL法筛选行人风险"源"因素，从总体上快速评估信号交叉口行人安全水平；微观上，划分行人冲突风险度时空计算单元，采用事件树与蒙特卡罗法计算风险度，分析行人风险时空分布特征，实现行人过街风险全过程动态评估，从而把握不同类型行人个体的安全水平。该研究成果可为从交通规划设计、运行管理的源头上消除或降低行人风险，实现主动风险防范提供理论支持和技术依据。

行人安全是交通安全领域的主要问题之一。我国大部分交叉口行人、机动车交互行为复杂多样，采取均一的冲突指标可能难以准确识别所有冲突。此外，作为交通参与者中的弱势群体，一旦发生事故往往导致行人非死即伤的严重后果，因此单纯的考虑冲突数量难以全面评价行人安全。因此，本研究提取了近万条上海市典型交叉口行人、机动车轨迹数据，从行为分析的角度入手深入挖掘行人冲突安全机理，根据交通参与者行为特征和冲突参数TTC、GT变化规律提出三类人车交互模式，即压迫交互、自由交互和协作交互，在此基础上建立了一整套“轨迹提取-交互行为模式-冲突指标”的人车冲突识别方法；引入风险分析理论，综合考虑人车碰撞概率和碰撞后果，建立了行人风险度计算方法、基于时空单元划分的行人时空风险地图分析方法、以及基于行人过街安全感受的行人风险分级标准。此外，结合国外行人事故数据和国内交叉口运行的实际特点，建立了行人宏观风险因素库。本项目的研究成果表明，基于行为的人车冲突识别方法可以更加准确的识别我国复杂条件下的人车冲突，时空风险地图方法则突破了传统的宏观整体的安全评价，考虑不同时空范围内的行人安全，全面诊断行人安全问题。因此，本项目可为从交通规划设计、运行管理的源头上消除或降低行人风险，实现主动风险防范提供理论支持和技术依据。