山区高速公路驾驶人加减速行为建模

山区地形地质条件复杂，各类复杂的组合线形设计更为常见，例如直线与平曲线间组合或不同平曲线间组合。驾驶人在相邻组合路段行驶时会感知到线形的变化，引起驾驶行为的改变，最终车辆的纵向加速度也会随之改变。频繁的加减速行为会引起驾驶人不适，甚至形成安全隐患。目前针对相邻组合路段驾驶行为的研究中，关于加速度的研究主要基于路段特殊点进行计算。随着驾驶模拟技术的发展，高仿真驾驶模拟器为高速公路的设计评估提供了更好的数据及试验条件支撑。在高仿真驾驶模拟器中，基于湖南省永吉高速公路道路设计参数及周边地形环境参数，构建山区高速公路的三维虚拟模型，以山区高速公路中的相邻组合路段为研究对象，获取山区高速公路组合线形路段的车辆纵向加速度数据，提取加减速事件后，基于驾驶人的加减速行为，采用混合Logit模型，分别判定道路线形层和驾驶人层的影响，研究组合线形对驾驶人纵向加减速选择的具体影响变量以及变量的影响范围。研究结果表明：下游路段最大曲率、上游路段圆曲线段比例、下游路段变坡点数量、下游路段曲线数量、上游路段平均曲率和当前位置曲率等对驾驶人加减速行为有显著影响；通过对比混合Logit模型和多元Logit模型，指出驾驶人层面对模型结果的影响显著。研究结果提供了一种山区高速公路连续纵向加减速行为的建模方案，并可为研究驾驶人在复杂线形条件下的纵向加速度选择行为提供基础。

Modeling of Driver Acceleration and Deceleration Behavior in Mountain Freeways

Mountainous terrain is complex, and various complex combination geometric designs are common, such as a combination of straight and flat curves or a combination of different flat curves. Drivers perceive changes in the line shape when driving on the adjacent combination section and change their driving behavior, thereby changing the longitudinal acceleration of the vehicle. Frequent acceleration and deceleration behavior can cause driver discomfort and even pose a safety hazard. In the current research on the driving behavior of adjacent combined sections, the research on acceleration is mainly based on special points of the road section. With the development of driving simulation technology, a high-simulation driving simulator provides good data and experimental conditions for freeway design. A three-dimensional virtual model was constructed in the driving simulator, based on the road design parameters of Hunan Yongji Freeways and the surrounding terrain environmental parameters. Taking the adjacent combined sections as the research object, the longitudinal acceleration data of a vehicle on a mountainous freeway combined linear section was obtained. Then, based on the driver's acceleration and deceleration behavior, a mixed Logit model was used to determine the influence of the road linear layer and the driver layer, respectively, as well as the influence parameters and influence range of the driver's longitudinal acceleration and deceleration behavior. The results show that the maximum curvature of the downstream section, the proportion of the circular section of the upstream section, the number of slope points of the downstream section, the number of curves of the downstream section, the average curvature of the upstream section, and the curvature of the current position have a significant impact on the driver's acceleration and deceleration behavior. Comparing the differences between the mixed and multivariate Logit models, the results show that the driver's impact on the model results is significant. This research provides a modeling scheme for continuous longitudinal acceleration and deceleration behavior of mountainous freeways, and provides a basis for studying the longitudinal acceleration selection behavior of drivers under complex linear conditions.