结合卡尔曼滤波的城市路段速度估计  

Urban link speed estimation with Kalman filter

作　　者：曾伟良[1] 何兆成[1] 沙志仁[1] 佘锡伟[1] 

机构地区：[1]中山大学智能交通研究中心广东省智能交通系统重点实验室,广州510275

出　　处：《测绘科学》2013年第1期96-99,共4页Science of Surveying and Mapping

基　　金：广东省科技计划资助项目(2009A011601013);广东省交通信息公众服务平台项目(GDIID2008IS006)

摘　　要：本文以城市出租车为浮动车数据采集源,介绍了基于GPS数据的实时路段速度估计的基本方法。针对目标路段GPS数据样本量不足的情况,考虑邻近区域的路段速度、上周同日速度、前一时刻速度等与目标路段当前时刻速度等密切相关的变量,建立多元线性回归方程,利用卡尔曼滤波融合预测值和测量值,从而提高路段行驶速度的估计精度。选择广州市东风路作为测试实例,融合值比测量值误差降低9%,绝对相对误差变动系数减少4%,表明结合卡尔曼滤波技术的城市路段速度估计精度和稳定性均得到提高。Regarded taxi as the source of probe vehicle data, a basic approach of urban real-time link speed estimation based on GPS data was introduced. Aimed at the condition of insufficient samples of GPS data, a multi-linear regression model was established by considering several variables such as the speeds of neighboring links, the speed of the objective link last week, and the previous speed of the objective link and so on, which closely relate to the real-time speed of the objective link. In order to improve the estima- tion precision, Kalman filter was adopted to merge the measured value and the predicted value. A case study of Dongfeng road in Guan- gzhou city was selected to test the proposed method and the experimental result indicated that the mean absolute percent error reduced 9% and the coefficient of absolute percent error alteration reduced 4%.

关 键 词：智能交通系统 浮动车 速度估计 卡尔曼滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] P228.4[自动化与计算机技术—计算机科学与技术]