【作者】 张源；

【导师】 谷远利；

【作者基本信息】 北京交通大学 ， 交通运输规划与管理， 2019， 硕士

【摘要】 城市快速路作为城市各个区域连接的大动脉,具有快速、高效、舒适的特点。为了更好的智能化的掌握了解城市交通运行规律、交通状态变化情况、缓解城市道路交通压力和为多方提供决策支持,研究城市快速路交通运行状态的判别和预测具有重要的意义。本文针对道路交通特性,从交通参数预测和交通状态判别两方面展开研究,本文主要研究的方面为:第一,归纳总结国内外交通状态判别划分、交通参数预测以及交通状态变迁演化三个方面的研究现状,掌握多种研究方法,并且确定本文的技术路线、章节安排和研究内容。第二,基于北京城市二环快速路的微波交通检测器数据,对交通流三参数之间关系模型进行了回顾分析,并且根据采集得到的交通数据对交通流三参数进行了时空特性分析,全面分析产生了这种现象的具体原因。第三,针对本文研究的交通状态变化的的情况,采用模糊c-均值聚类算法将交通三参数两两组合作为数据输入,得到聚类中心和交通状态判别划分信息;标定交通状态判别精度,根据精度确定最佳参数组合,通过实例验证得出流量-速度为判别交通状态的最佳参数组合。第四,为了更好地研究交通状态变化,进行参数预测,为了保证预测精度,采用免疫算法优化的最小二乘支持向量机(LSSVM)建立交通流量预测模型。通过免疫算法优化经过训练后的LSSVM中的惩罚因子和核函数参数,得到最优的预测模型。将行驶速度和占有率作为模型的输入,交通流量作为输出。第五,为了更好地展现交通状态判别情况以及交通状态演化情况,本文应用有限状态机模型作为模型。将预测后的交通参数,输入到有限状态机进行交通状态判别和演化。实验结果证明,有限状态机更为直观地展现了交通状态演化情况,也证明了有限状态机在交通状态判别演化中的可用性,能够更好地为城市道路交通管理与控制服务。更多还原

【Abstract】 Urban expressway,as the main artery connecting various areas of the city,has the characteristics of fast,efficient and comfortable.In order to better understand the city intelligent traffic law,traffic state changes,alleviate the city traffic pressure and support for the parties providing the decision-making,judgment and has important significance for prediction of city expressway traffic running state.In this paper,the characteristics of road traffic are studied from two aspects:traffic parameter prediction and traffic state discrimination.The main research aspects of this paper are as follows:Firstly,this paper summarizes the research status of traffic state classification and discrimination,traffic parameter prediction and traffic state evolution at home and abroad,grasps a variety of research methods,and determines the technical route,chapter arrangement and research content of this paper.Secondly,based on the microwave traffic detector data of Beijing Second Ring Expressway,the relationship model between the three parameters of traffic flow is retrospectively analyzed,and the spatial and temporal characteristics of the three parameters of traffic flow are analyzed according to the collected traffic data,and the specific reasons for this phenomenon are comprehensively analyzed.Thirdly,in view of the traffic state changes studied in this paper,the fuzzy c-means clustering algorithm is used to input two groups of traffic three parameters into the data to obtain the clustering center and traffic state partition information;the accuracy of traffic state discrimination is calibrated,and the best combination of parameters is detemiined according to the accuracy,and the flow-speed is the best combination of parameters through the example verification.Fourthly,in order to better study the changes of traffic conditions and predict traffic parameters,and to ensure the accuracy of traffic parameters prediction,the least squares support vector machine(LSSVM)optimized by immune algorithm is used to establish traffic flow prediction model.The immune algorithm is used to optimize the penalty factor and the parameters of the kernel function in the trained LSSVM,and the optimal prediction model is obtained.Velocity and occupancy are taken as input and traffic flow as output.Fifthly,in order to better show the traffic state discrimination and the traffic state evolution,this paper uses the finite state machine model as the model.The predicted traffic parameters are input into the finite state machine to discriminate and evolve the traffic state.The experimental results show that the finite state machine can more intuitively show the evolution of traffic state,and also prove the availability of the finite state machine in the evolution of traffic state discrimination,which can better serve the urban road traffic management and control.更多还原