基于贝叶斯网络的车辆换道决策模型研究  

Decision Model of Vehicle Lane Change Based on Bayesian Network

作　　者：赵树恩[1] 柯涛 柳平 ZHAO Shuen;KE Tao;LIU Ping(College of Mechatronics&Vehicle Engineering,Chongqing Jiaotong University,Chongqing 400074,China)

机构地区：[1]重庆交通大学机电与车辆工程学院,重庆400074

出　　处：《重庆交通大学学报（自然科学版）》2020年第5期130-137,144,共9页Journal of Chongqing Jiaotong University(Natural Science)

基　　金：国家重点研发计划项目(2016YFB0100905);重庆市基础研究与前沿探索项目(cstc2018jcyjAX0422);中国博士后基金项目(2014M562259);重庆市博士后基金项目(XM2014084)。

摘　　要：车辆换道行为是微观交通流中的典型驾驶行为之一。车辆换道决策模型研究可以为自动驾驶汽车协同自适应巡航控制(cooperative adaptive cruise control,CACC)提供理论基础,也能有效减少车辆危险换道行为引发的交通事故。为使换道模型更加适应动态道路交通环境,以美国交通部联邦公路管理局NGSIM项目实测试验数据为依据,分析车辆换道决策时自身车辆速度、加速度及其与交互车辆相对时距等相关特征参数,并运用贝叶斯网络人工智能理论,建立车辆换道决策模型,通过仿真分析并与NGSIM实测数据进行对比。结果表明:基于贝叶斯网络的换道决策模型的平均决策准确度和识别率可达到89%以上,具有良好的换道决策效果,可为智能车辆协同自适应巡航控制及自动驾驶深度学习提供理论参考。Lane change behavior is one of the typical driving behaviors in micro traffic flow. Researching vehicle lane change decision-making model can provide a theoretical basis for cooperative adaptive cruise control( CACC) of automatic driving vehicles,and can also effectively reduce the traffic accidents caused by dangerous lane-changing behavior. To make the lane change model more adaptable to the dynamic traffic environment,according to the measured test data of NGSIM project of Federal Highway Administration of the U.S. Department of Transportation,the relevant characteristic parameters such as own vehicle speed,acceleration and relative time distance between vehicles and interactive vehicles during the decision-making of vehicle lane change were analyzed. And by using the theory of Bayesian network artificial intelligence,the decisionmaking model of vehicle lane change was established,which was compared with the measured data of NGSIM through simulation analysis. The results show that the average decision-making accuracy and recognition rate of the lane change decision-making model based on Bayesian network can reach more than 89%,which has good lane change decision-making effect,and can provide theoretical reference for CACC of intelligent vehicles and depth learning of automatic driving.

关 键 词：车辆工程 智能交通 微观交通流 车辆换道 贝叶斯网络 

分 类 号：U491[交通运输工程—交通运输规划与管理]