【作者】 张洋；

【导师】 张惠玲；

【作者基本信息】 重庆交通大学 ， 交通信息工程及控制， 2018， 硕士

【摘要】 随着社会经济的发展,小汽车拥有量的持续增长,交通基础设施建设的完善,道路行驶安全受到严重威胁,交通事故更是频繁发生,尤以信号交叉口车辆冲突最为严重。其中,两难区现象使得车辆在通过交叉口时难以确定是否通过,加剧交通矛盾,影响驾驶员安全。研究驾驶员行驶特性,解决城市道路两难区问题,缓解交叉口车辆冲突矛盾,提高道路通行能力是目前亟待解决的重要课题。基于此,本文以两难区中的车辆作为研究对象,从两难区影响外因与驾驶员选择行为内因两方面入手,在对数据统计分析的基础上,基于Logit和HMM分别构建了两难区驾驶员选择行为模型。本论文的主要研究内容包括以下几个方面:(1)采用RP调查与SP调查相结合的方法,设计调查问卷并对城市道路的车辆进行视频拍摄,提取现场车辆自由速度,通过Vissim软件在不同车道以及不同公交比例条件下进行仿真,结果发现车道以及公交比例对两难区的长度存在着显著影响。(2)基于SEM模型,对动态两难区的内因做深层次的研究,研究结果表明:驾驶员的驾驶信息对于驾驶选择行为有显著影响且为负相关,其次是社会属性对于驾驶行为的影响,关于辅助系统的意见对驾驶选择行为有影响但影响最弱,其效应值分别为0.53、0.41、0.35。(3)根据SEM模型的结论,选取驾驶信息以及社会属性指标构建了基于二元Logit的驾驶员选择行为模型,结果发现:驾龄、驾照类型以及车道是影响驾驶员选择概率的主要因素;驾龄大的采取停车概率较大,驾照类型为C证的更倾向于停车,而在最右侧车道的驾驶员加速通过的概率更大。通过实际调查与模型的预测值对比发现,有倒计时的大路口下,模型准确率为66.67%。(4)本文采用隐马尔科夫模型,输入观测序列距离交叉口位置、绿灯倒计时、车道、前方是否有车四项观测序列信息,分别构建了加速通过以及减速停车两种状态识别模型,并验证模型识别精度达到67%。更多还原

【Abstract】 With the development of society and economy,the continued growth of car ownership,the improvement of transportation infrastructure,the serious threat to road safety,and the frequent occurrence of traffic accidents,especially at the signalized intersections.Among them,the dilemma phenomenon makes it difficult for the vehicles to pass through the intersection,whether it passes or not,intensifies the traffic contradiction,and affects the safety of the driver.Studying the driver’s driving characteristics,solving the problem of the urban road dilemma,alleviating the contradiction between vehicles at the intersection,and improving the road traffic capacity are important issues to be solved urgently.Based on this,this paper takes the vehicle in the dilemma area as the research object,starting with the influencing factors of the dilemma area and the driver’s choice behavior.Based on the statistical analysis of the data,the driver’s selection behavior of the dilemma area is constructed based on Logit and HMM respectively.model.The main research contents of this paper include the following aspects:(1)Using a combination of RP survey and SP survey,the survey questionnaire was designed and the vehicles on the urban roads were video-captured.The free speed of the vehicles on the site was extracted.Simulations were performed with Vissim software in different lanes and different bus ratios.There is a significant effect on the length of the dilemma.(2)Based on the SEM model,the driver’s choice of behavioral mechanism is studied in depth.The results show that the driver’s driving information has a significant influence on the driving choice behavior and is positively correlated,followed by the impact of social attributes on driving behavior,and on the auxiliary system.The opinions on the choice of driving behavior have an impact but the weakest effect.Their effect values are 0.53,0.41,and 0.35,respectively.(3)According to the conclusion of the SEM model,driving information and social attributes were selected to construct a driver selection behavior model based on binary Logit.The results showed that the driving age,driver’s license type,and lane are the main factors affecting thedriver’s selection probability;The parking probability is greater,the driver’s license type is more likely to stop,while the driver in the right-most lane has a higher probability of accelerating.Through the actual investigation and the predicted value of the model,it is found that the accuracy of the model is 66.67% under the countdown intersection.(4)In this paper,the hidden Markov model is used to input the observation sequence from the intersection location,the green light countdown,the lane,the front of the vehicle there are four observation sequence information,respectively to build the acceleration pass and deceleration stop two state recognition models,and verify the model recognition accuracy It reaches 67%.更多还原