近年来，随着我国交通需求和公路里程的不断增长，公路上的交通事故也随之增加。现研究表明特殊路段处的事故率明显高于正常路段，对此，道路交通管理者采取了诸多治理措施，并取得了一定的成效。但是，从某种程度上讲，发生道路交通事故的主要原因是驾驶员没能及时发现危险并采取规避措施。对此，论文提出了通过建立基于道路交通设施主动安全的预警系统来预防公路特殊路段事故发生的研究思路，并对此进行了深入的研究。 首先，论文研究分析了当前可用于公路特殊路段的交通信息采集技术和气象信息采集技术，为特殊路段预警系统提供数据支撑。 接着，论文分析了车辆在无信号T型交叉口和小半径曲线路段处的速度特征，并给出了车辆速度变化点的统计值。建立了广义回归神经网络速度预测模型，并给出了模型的自我评价标准和自我更新流程。建立了二次多项式速度预测模型，并与广义回归神经网络速度预测模型进行对比，发现广义神经网络速度预测模型擅长于预测车辆速度和行驶时间，而二次多项式速度预测模型擅长于预测车辆速度。 通过分析影响公路特殊路段车辆行车安全的交通因素、气象因素、道路线形因素，建立了有交通冲突特殊路段的危险判别模型和无交通冲突路段的危险判别模型。并依据危险判别模型的计算结果对车辆的行车安全等级进行分类——安全、较安全、较危险、危险、高度危险。 论文以交通行为与交通安全驾驶模拟器为实验平台，研究了10种不同的警示设施在无信号交叉口处对驾驶员驾驶行为的影响，得出以下结论：1、标志牌类警示设施宜布设在距离交叉口10米处，警示灯类设施宜布设在交叉口处；2、警示灯的闪烁频率越高，给驾驶员的心理压力越大，但是对驾驶员驾驶行为的影响不一定越大。3、“标志牌+警示灯”类预警设施更能有效的使驾驶员提前减速，延长减速距离。 最后，论文构建了适用于公路特殊路段的预警系统，并以无信号T型交叉口和小半径曲线为例进行分析说明。 关键词：道路交通设施，主动安全，速度预测模型，危险判别模型，警示设施有效性 |
In recent years, with the development of the increase of traffic demand and road mileage in China, the amount of traffic accident grows rapidly. Existing studies shows that the special sections of road have a higher accident rate than normal sections. Many effective measures have been used to decrease traffic accident on special sections. However, to some extent, the main reason of road traffic accidents is that the driver was not able to detect danger and take evasive measures. This paper proposes a pre-warning system based on traffic facility active safety to minimize the accident in special sections. Firstly, the traffic information collection methods and meteorological information collection technology available for road special sections is summarized and analyzed. Secondly, the character of vehicle speed is analyzed in no signal T-shaped intersection and small radius curve road, and statistical data of the vehicle speed change points are given. This paper build up a generalized regression neural network speed prediction model, and gives self-evaluation standard and self-renewal process of the model. This paper establish a quadratic polynomial speed prediction models, and with the general regression neural network prediction model to compare and found that generalized neural network prediction model is good at predicting vehicle speed and travel time, and quadratic polynomial prediction model is good at predicting vehicle speed. Thirdly, the main effect factors of traffic safety on road special sections are traffic, meteorological environment and road alignment. Based on the analysis of these effect factors, two special section danger discriminant models with traffic conflict or not are proposed. According to the calculating results of the danger discriminant models, the road special section safety can be classified into very safe, safe, dangerous, and very dangerous. And then, based on the driving simulator platform, 11 traffic warning facilities for unsignalized intersections are chosen to study the impact of different warning facilities on driving behavior. Some conclusions can be made. First, signboard warning facilities have the best performance when they are set 10m to the intersection, and warning light facilities should be set in the intersection. Second, the higher frequency the warning light flashes, the more mental stress expresses to the drivers, but not certainly more impact on the driving behavior. Third, the combination use of signboard and warning light can well make the driver speed down and increase the deceleration distance. Finally, on the base of all the researches above, a pre-warning system for road special sections is structured, and illustrated by the example of no signal T-shaped intersection and small radius curve road. Key Words: road traffic facility, active safety, speed prediction model, danger discriminant models, validity of warning facility |
硕士论文