基于收费数据的高速公路旅行时间自适应插值卡尔曼滤波预测研究

发布时间：2018-12-18 17:37

【摘要】：摘要：高速公路站间旅行时间可衡量所经路段通行效率和交通状态,是交管部门交通控制和诱导的重要依据,也是出行者高度关注的首要信息,已成为先进出行者信息系统ATIS (advanced traveler information systems)和路径导航系统RGS (route guidance systems)的关键因素。本文依托高速公路收费数据开展站间旅行时间预测研究,具体研究内容如下： (1)针对MTC(manual toll collection)数据相比ETC (electronic toll collection)数据包含车辆排队等待缴费时间的问题,提出一套MTC和ETC数据实时融合处理准则,包含极端异常数据处理和数据融合,提高了周期内车辆数据数量。 (2)针对收费数据中异常数据难以剔除的问题,提出一种改进后平均旅行时间计算模型,模型融入四分法数据剔除思想,提高了收费数据质量和平均旅行时间计算精度。 (3)针对卡尔曼滤波算法非线性性能弱及自适应性能差的问题,提出了高速公路旅行时间自适应插值卡尔曼滤波算法。算法利用等间距插值方法重构实时及历史旅行时间之间的时间序列,基于最小二乘法实时搭建卡尔曼滤波模型,并详细阐述了Sage-Husa自适应卡尔曼滤波旅行时间预测原理。 (4)为验证算法有效性,实际路段算法验证结果可知,自适应插值预测算法在正常、事故、小长假三种交通流状态下所有周期平均相对误差控制在7.5%内,事故周期平均相对误差控制在10%内。 (5)搭建了旅行时间预测系统架构和预测系统业务逻辑,详细阐述了预测系统收费数据存储数据库设计、旅行时间算法设计、发布界面设计,并基于C#. SQL Server2008开发了高速公路站间旅行时间预测系统。 (6)搭建了离线旅行时间系统稳定性测试环境,系统运行稳定后,将其布设到高速公路信息中心,实时预测高速公路旅行时间。京港澳高速公路旅行时间预测示范系统应用良好,可为公众出行提供时间参考。
[Abstract]:Absrtact: Expressway interstation travel time can measure the traffic efficiency and traffic state, which is the important basis for traffic control and guidance of traffic management department, and is also the most important information that travelers pay close attention to. It has become the key factor of the advanced traveler information system (ATIS (advanced traveler information systems) and the path navigation system (RGS (route guidance systems). Based on highway toll data, this paper carries out the research of interstation travel time prediction. The specific research contents are as follows: (1) compared with ETC (electronic toll collection) data, MTC (manual toll collection) data contains the problem of vehicle queuing waiting for payment time. A set of real-time fusion criteria for MTC and ETC data is proposed, which includes extreme anomaly data processing and data fusion, which improves the number of vehicle data in the cycle. (2) aiming at the difficulty of eliminating abnormal data in toll data, an improved average travel time calculation model is put forward. The model integrates the idea of quaternion data elimination, and improves the quality of toll data and the accuracy of average travel time calculation. (3) aiming at the problem of weak nonlinear performance and poor adaptive performance of Kalman filtering algorithm, an adaptive interpolation Kalman filter algorithm for expressway travel time is proposed. The algorithm uses equal-space interpolation method to reconstruct the time series between real time and historical travel time. Based on the least square method, the Kalman filter model is built in real time, and the principle of Sage-Husa adaptive Kalman filter travel time prediction is described in detail. (4) in order to verify the validity of the algorithm, the experimental results show that the average relative error of all periods is controlled within 7.5% under normal traffic flow, accident and small length false traffic flow. The average relative error of accident period is controlled within 10%. (5) the structure of the travel time prediction system and the business logic of the prediction system are built. The design of the database of charge data storage, the design of travel time algorithm, the design of the publishing interface, and the design based on C#are described in detail. SQL Server2008 has developed a travel time prediction system between freeway stations. (6) an off-line travel time system stability test environment is built. After the system runs stably, it is arranged in the expressway information center to predict the expressway travel time in real time. The Beijing, Hong Kong and Macao Expressway Travel time Prediction demonstration system is applied well and can provide time reference for public travel.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495

【参考文献】