摘要
为研究驾驶员的跟车特性,探究驾驶员风格划分方法,采集50名驾驶员的实车试验数据,选取平均跟车时距和平均制动时距为二维向量,建立基于K-means聚类结果的高斯混合模型(GMM)并分析不同风格驾驶员的聚类结果。研究表明:样本数据聚为3类时的平均轮廓系数为0. 45,将驾驶员划分为冒进型、平稳型、保守型3类时聚类效果较好;冒进型驾驶员倾向于选择较小的跟车时距和制动时距,保守型驾驶员的跟车及制动时距则普遍较大,模型聚类结果更加柔性,样本区分度更高。
In order to study drivers ’ car-following characteristics and explore an effective method to classify driving styles,50 participants were recruited to carry out a real road driving test. A GMM with results of K-means clustering was established based on two-dimensional variables: average car-following time gap and average braking time gap. And then results of different types of drivers were analyzed. The research shows that clustering result is better with three categories( aggressive drivers,steady drivers,and conservative drivers) with an average contour value of 0. 45. It is found that aggressive drivers tend to choose a smaller car-following time gap or braking time gap while conservative drivers usually take a larger value,and a much softer clustering result with a high separability between samples would be achieved.
作者
刘通
付锐
张名芳
田顺
LIU Tong;FU Rui;ZHANG Mingfang;TIAN Shun(School of Automobile,Chang'an University,Xi'an Shaanxi 710064,China;Key Laboratory of Automobile Transportation Safety Technology,Ministry of Transport,Chang'an University,Xi'an Shaanxi 710064,China;School of Electrical and Control Engineering,North China University of Technology,Beijing 100144,China)
出处
《中国安全科学学报》
CAS
CSCD
北大核心
2019年第12期40-45,共6页
China Safety Science Journal
基金
国家自然科学基金资助(51775053)
国家重点研发计划(2018YFB1600500)
教育部长江学者与创新团队发展计划项目(IRT_17R95)
汽车运输安全保障技术交通行业重点试验室开放课题(300102229505)。
