基于驾驶员视觉特性的沙漠草原公路长直线段对行车安全影响研究

According to traffic accident data, the influence of the long straight line segment on driving safety is unignorable. Especially for desert steppe area, highway design often use straight segment because of its' gentle slope.But traffic accidents will be caused by long straight line segment more easily because of the monotonous landscape of desert steppe region. In China, "Design Specification for Highway Alignment" fail to makes clear definition for the maximum straight line length on the highway. Aiming to the effect of long straight section of desert steppe road on drivers visual characteristics ,which including visual search features, visual information load and visual fatigue, this research was carried through on the basis of analyzing data collected in drivers' driving tests as well as models building. Firstly , drivers' eye movement data and driving data are obtained by driving tests.Secondly, modeling of "visual interest areas" and "visual information content" are built.Thirdly ,the changes of "unit time visual interest area" and relation model between "pupil diameter approximate entropy" and "visual information content" are be analyzed in the microscopic scales. Finally, the changes of "unit time visual interest area" index and relation model between "pupil diameter approximate entropy" and "visual information content" with time accumulation and alignment condition accumulation are be analyzed in the macroscopic scales. Based on the analysis above, the influence mechanism of long straight line segment on visual search features, visual information load and visual fatigue can be obtained, and corresponding values of stright segment length for desert steppe highway can be confirmed using extreme value theorem and Lagrange mean value theorem.Above all ,this research can provides evidence for highway alignment design.

项目针对沙漠草原公路长直线路事故率高且我国《公路路线设计标准》未对直线段最大长度明确界定的实际，以数据分析和建模为手段，考虑驾驶员在行车过程中视觉搜索、视觉信息负荷和累积视觉疲劳影响，通过对行车过程眼动参数的采集、"视觉兴趣区"模型和"视觉信息量"模型的建立，首先分析"单位时间视觉兴趣面积"和"单位时间视觉兴趣区重心坐标"微观变化，确定长直线段对驾驶员视觉搜索稳定性影响机理；其次，分析"瞳孔直径近似熵"与"视觉信息量"关系模型，确定长直线段对驾驶员视觉信息负荷影响机理；再次，确定随时间累积和线形累积的"单位时间视觉兴趣区"指标变化和"瞳孔直径近似熵"与"视觉信息量"关系模型变化，确定长直线段对视觉疲劳累积的影响机理；最后，通过应用极值定理和拉格朗日中值定理对相关模型分析，确定公路长直线段长度、后接平曲线半径和直线段线形比例和分布，为沙漠草原公路线指标的修订和公路路线安全评价标准提供依据。

课题研究以数据分析和建模为手段，考虑驾驶员在行车过程中视觉搜索、视觉信息负荷和累积视觉疲劳影响，通过对宁夏境内国道110、京藏高速（宁夏段）、定武高速、青银高速4条道路等4条公路上展开了总测试里程超过八万公里的驾驶员行车试验，采集了驾驶员行车过程中的眼动参数、建立了“视觉兴趣区”模型和“视觉信息量”模型。（1）通过分析“单位时间视觉兴趣面积”和“单位时间视觉兴趣区重心坐标”微观变化，确定长直线段对驾驶员视觉搜索稳定性影响机理，从驾驶员视觉兴趣搜索角度确定理想的长直线行车时间为60秒，当直线段行车时间超过140秒，驾驶员不再进行积极的视觉搜索行为；（2）通过分析“瞳孔直径近似熵”与“视觉信息量”关系模型，确定长直线段对驾驶员视觉信息负荷影响机理，确定长直线段驾驶员视觉信息可承载的行驶车速阈值为124km/h，当高速行驶状态大于81秒后，驾驶员将适应高速行驶状态，这将导致一旦有危险将来不及做出正确反应；（3）时间累计效应下对驾驶员视觉疲劳的影响，建立宏观的时间累积效应下长直线段驾驶员视觉疲劳模型，确定长直线段对视觉疲劳累积的影响机理；（4）通过应用极值定理和拉格朗日中值定理对相关模型分析，确定公路长直线段长度、后接平曲线半径和直线段线形比例和分布，为沙漠草原公路线线形指标的修订和公路路线安全评价标准提供依据。