摘要
基于上海延安高架两个驶入匝道(南线虹许路驶入匝道、虹井路驶入匝道)瓶颈的汇入行为视频,对汇入车辆、当前和目标车道汇入交互车辆进行了轨迹和汇入行为参数提取,共获得416个汇入行为样本;应用分类回归树(CART)对3种汇入行为分别进行建模,分析影响不同汇入行为的因素,并用混淆矩阵对分类结果进行评价.结果表明,CART能较好地预测3种不同汇入类型,其分类准确率均达到了75%以上.CART与经典离散选择模型和朴素贝叶斯分类结果对比表明,CART的分类效果明显优于上述两类模型.
Based on the videos of traffic flow at two bottlenecks (Hongxu on-ramp and Hongjing on-ramp) on Yan' an Expressway in Shanghai, 416 empirical merging behavior samples were collected by extracting trajectories from merging vehicles, as well as each adjacent vehicles. The classification and regression tree (CART) was adopted for modeling three merging situations, the key parameters affecting different merging behaviors were analyzed and the confusion matrix was used to evaluate the result of the classification accuracy. The results show that CART performed well with these data. All the accuracies are over 75%. Moreover, a comparison among CART, classical discrete choice model and naive Bayes classifier was conducted, and the CART shows the best classification results.
出处
《同济大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2015年第4期549-554,共6页
Journal of Tongji University:Natural Science
基金
国家自然科学基金(51278362
51422812)
关键词
城市快速路
驶入匝道瓶颈
汇入行为
分类回归树
交通流失效
urban expressway
on-ramp bottleneck
merging behaviors
classification and regression tree
trafficbreakdown