车联网中驾驶员反应时间实时估计方法

摘要

结合车辆非线性分段制动特性和车辆状态数据对驾驶员反应时间进行实时估计，采用特征数据拟合出初速度与减速度上升时间关系模型，通过BP神经网络推算减速度上升过程中减速度随时间非线性变化关键参数，并在此基础上基于车辆动力学运动规律求解车辆分段制动位移，通过实际位移与理论位移等价关系实现驾驶员反应时间实时估计，最后利用Newell跟驰模型场景对本文提出方法进行仿真验证.结果表明：不同初速度条件下，反应时间估计值平均误差不超过0.022 s；不同反应时间条件下，反应时间估计值平均误差不超过0.013 s.本文提出的反应时间估计方法不用增加通信负担和额外设备，能够达到较高精度，具备更优的适应性、可用性及工程价值.%A driver reacting time real-time estimating method based on nonlinear piecewise braking features and vehicle status is proposed in this paper. Numerical fitting approach is used to illuminate the relationship of initial velocity and deceleration rising time. And a BP neural network is established to model the main parameters of the a-t curve. Based on the two vehicle dynamic models, it estimates the driver reacting time through comparison of the theoretical displacement and the actual displacement. It sets up a Newell car following model based simulating environment to verify the methodology. The results show that under the condition of different initial velocity, the error of estimated reacting time is less than 0.022 s. And under the condition of different reacting time, the error of estimated reacting time is less than 0.013 s. The method proposed in the paper doesn’t need extra information interaction and measurement devices. And the precision can meet the requirement of vehicle safety. It can be used in traffic engineering.