地下快速路驾驶人速度感知敏感性分析及模型构建

Analysis and Modeling of Speed Perception Sensitivity of Drivers on Underground Expressways

摘要:

为探究驾驶人速度感知机理，设计驾驶模拟实验采集驾驶人感知速度与实际行车速度，分析地下快速路几何线形、侧壁变化频率等因素对驾驶人速度感知敏感性的影响，并基于恒定速度理论、机器学习方法构建驾驶人速度感知偏差敏感性阈值预测模型。结果表明，在0.05显著水平下，不同速度水平、线形组合、侧壁变化频率组别下，驾驶人的速度感知敏感性阈值都存在显著差异。采用LASSO（least absolute shrinkage and selection operator）筛选根据短期、中长期、长期时窗划分的各特征变量，构建的多元非线性模型的决定系数R2为0.645，多层感知机模型的决定系数R2为0.727，支持向量回归模型的决定系数R2为0.853。

Abstract:

To explore the mechanism of the speed perception of the driver， a driving simulation experiment was designed to collect the perceived speed and actual driving speed of the driver，which analyzes the influencing factors such as the geometric alignment of the underground expressway and the side wall change frequency on the speed perception sensitivity of the driver. The prediction model of the speed perception bias sensitivity threshold of the driver was constructed based on the constant speed theory and the machine learning method. The results show that at the 0.05 significant level， there are substantial differences in the speed perception sensitivity threshold of the driver at different speeds ， linear combinations， and sidewall change frequency groups. The Lasso variable filters the characteristic variables according to the short-term， mid-long-term， and long-term time windows. The determination coefficient R² of the constructed multivariate nonlinear model is 0.645， the determination coefficient R² of the multilayer perceptron model is 0.727， and the determination coefficient R² of the support vector regression model is 0.853.