基于数据融合的超速事件时空计量模型构建及协同干预方法研究

Speeding event increases traffic accident occurrence probability and its severity. As for speeding event, some intervention measures have been successively proposed and implemented. They have obtained a certain effect. However, this kind of traffic offence is still very common. The reason is that intervention approaches for the speeding event are less systematic and lack theoretical basis. In view of such problem, this project fuses vehicle GPS trajectory data, speeding offence records, and driving experiment data. After that, by treating the speeding event as a line event, the detection algorithms for speeding event based on the continuous overspeed GPS trajectory points and average speed are established respectively. And on this basis the present research further fuses the detected speeding events, speeding offence records, road traffic environmental data, and traffic accident data. Based on the fused dataset, after defining the characteristic indexes of speeding event, the extraction approaches of its spatiotemporal distribution features are investigated. Given that the spatiotemporal correlation and variation characteristics of the influences of road traffic environmental factors on speeding event, spatiotemporal measurement models of speeding event are proposed respectively. According to the proposed models, an optimization model of the layout of speed camera is built in combination with traffic management and control measures development. In the meantime, a novel model of decreasing-block score for driving safety performance is established by considering the frequency of speeding events detected from the GPS trajectory data. By including the score, the contents and corresponding manners for informing the speeding driver are explored respectively. And thus, a synergy intervention method of speeding event is developed by simultaneously considering personalization and generality. The findings from this project provide the theoretical foundation for the long-term effective intervention of speeding event.

超速事件增加交通事故发生概率及其严重程度。对此，一些干预方法相继被提出且已实施，取得了一定的效果，但该交通违法现象仍然十分突出。究其原因是超速事件干预方法不系统且缺乏理论依据。鉴于此，本项目以融合车辆GPS轨迹数据、超速违法记录和行车实验数据为基础，将超速事件视为线事件，分别设计基于连续超速GPS轨迹点和平均速度的超速事件探测算法。在此基础上，融合经探测的超速事件数据、超速违法记录、道路交通环境数据和交通事故数据，界定超速事件特征指标，探究超速事件时空分布特征提取方法；考虑道路交通环境因素对超速事件影响的时空关联及变化特性，分别构建超速事件时空计量模型；据此构建速度照相机布设优化模型，提出交通管控对策；同时考虑经探测的超速事件频数，构建行驶安全绩效阶梯递减评分模型，研究超速事件告知内容及方式，从而提出兼顾个性化和共性化的超速事件协同干预方法。研究成果可为超速事件的长效干预提供理论依据。

超速事件增加交通事故发生概率及其严重程度。对此，一些干预方法相继被提出且已实施，取得了一定的效果，但该交通违法现象仍然十分突出。究其原因是超速事件干预方法不系统且缺乏理论依据。鉴于此，本项目以融合车辆GPS轨迹数据、超速违法记录和行车实验数据为基础，将超速事件视为线事件，分别设计基于连续超速GPS轨迹点和平均速度的超速事件探测算法。在此基础上，融合经探测的超速事件数据、超速违法记录、道路交通环境数据和交通事故数据，界定超速事件特征指标，探究超速事件时空分布特征提取方法；考虑道路交通环境因素对超速事件影响的时空关联及变化特性，分别构建超速事件时空计量模型；据此构建速度照相机布设优化模型，提出交通管控对策；同时考虑经探测的超速事件频数，构建行驶安全绩效阶梯递减评分模型，研究超速事件告知内容及方式，从而提出兼顾个性化和共性化的超速事件协同干预方法。结果表明，基于相邻多点间平均速度的超速事件探测算法精度最高，达到95%以上；超速事件具有空间正向聚集性且不随时间变化，低范围和中范围超速事件多以高-高局部空间模式呈现，低范围重复超速事件频仍；根据超速频次、严重度和持续时间，可将超速者划分为四类，其受运营因素的影响各异；考虑空间相关性、空间异质性和溢出效应的随机参数空间模型具有离散的残差、更多的跨0置信区间和更强的解释能力且更具时间稳定性，充分揭示各类因素的影响；基于阶梯递增处罚的个性化干预方法的干预效果稳定性更好；量化了路段和交叉口速度照相机威慑范围，确定某区域内速度照相机最佳个数和位置；兼顾个性化和共性化的协同干预方法效果最佳。研究成果可为超速事件的长效干预提供理论依据。

内容获取失败，请点击重试