快速路入口匝道车辆汇入决策模型

Decision-making Model for Vehicle Merging on Expressway On-ramp

摘要:

快速路合流区交通行为复杂，车辆行驶轨迹不固定，导致车辆行为难以预测。为了深入研究入口匝道车辆汇入决策发生条件，在不做任何理想化假设的情况下，对平峰、高峰期间快速路合流区汇入车辆行驶数据进行全区域采集。通过统计合流区车辆汇入间隙频率分布来分析汇入行为发生条件。发现高峰与平峰汇入间隙差别明显，初步判断外侧车道高峰临界间隙约为3 s，平峰临界间隙约为2 s。在此基础上，梳理高峰期间数据，筛选出对车辆汇入决策影响较重要的变量，建立高峰匝道车辆汇入决策Fisher判别[1]模型。最后，对所建立模型进行交叉验证，发现模型准确率可达87%。

Abstract:

The traffic behavior in expressway confluence area is complex and the trajectory of vehicle is not fixed, which make the vehicle behavior hard to predict. In order to further study the on-ramp vehicle merging decision-making conditions, under the condition of not making any idealized assumptions, the data of vehicles merging in the expressway confluence area during flat peak and peak periods are collected in whole area. The occurrence conditions of the merging behavior are analyzed by statistically analyzing the frequency distribution of the merging intervals of vehicles in the merging area. It is found that there is a significant interval difference between the peak and flat peak. The initial judgment is that the critical interval of the outside lane peak is about 3 s, and the critical interval of the flat peak is about 2 s. On this basis, the data during the peak period is sorted out, the variables that are more important to the decision-making of vehicle merging are screened out, and a Fisher discriminant model for the decision-making of vehicle merging on the peak ramp is established. Finally, the established model is cross-validated, which is found that the accuracy of the model can reach 87%.