交通流非参数回归模型  

Non-parameter Regression Model of Traffic Flow

作　　者：达庆东 段里仁 

机构地区：[1]北京市交通工程科学研究所,北京100061

出　　处：《数理统计与管理》2003年第4期41-46,共6页Journal of Applied Statistics and Management

摘　　要：交通流宏观参数流量、速度和占有率在交通工程和管理中具有非常重要的作用 ,对这三者关系的刻划反映了道路本身特性和交通流的规律。到目前为止 ,基本上采用线性或非线性的参数模型来描述 ,此类模型在应用中具有一定的局限性 ,并交通变量时间序列进行预测 ,数据拟合表明 ,选择适当的核函数或邻近数以及窗宽 。The traffic macro variables flow,speed and occupancy play an important role in the field of traffic engineering and management. By describing the relationship among these three variables, we can understand the characteristics of the road itself and the regularity of traffic flow. Until now, most often linear and non linear parameter models have been used to this domain,but with the limit somehow. This paper utilizes non parameter regression to bulid traffic flow occupancy,flow speed and speed occupancy models,and predicts traffic variables with the change of time; It's verified by data fitting that by selecting appropriate window width and number of neighbors,the results will be satisfactory.

关 键 词：交通流 非参数回归 模型 权函数 核估计 近邻型估计 时间序列预测 

分 类 号：U491.112[交通运输工程—交通运输规划与管理] O212.1[交通运输工程—道路与铁道工程]