基于高斯混合模型的车辆自由流速度分布  

Distribution of Vehicle Free Flow Speeds Based on Gaussian Mixture Model

作　　者：徐程[1] 

机构地区：[1]浙江警察学院治安系,浙江杭州310053

出　　处：《公路交通科技》2012年第8期132-135,158,共5页Journal of Highway and Transportation Research and Development

基　　金：浙江警察学院校级科研培育类项目(20110616)

摘　　要：基于车辆速度参数是描述交通流运行的最基本参数之一,在交通状态评价与交通安全性分析中具有的重要作用,针对城市道路不同种类车型混合运行的特征,提出采用高斯混合模型来描述自由流状态下的车辆速度分布。结合实测速度数据中的双峰分布特性,将混合车流车型分为大型车与小型车2类,进而建立了2类型的高斯混合模型,并采用期望最大化算法对模型参数进行估计。通过实测数据验证结果表明,模型能够很好的拟合观测速度数据,并能够有效区分不同类型车辆速度,为确定不同类型车辆限速标准、分析不同类型车辆相互影响提供理论依据。The vehicle speed parameter is one of the most fundamental parameters for describing the operation of traffic flow effectively. It is an important role in the traffic condition evaluation and analysis of traffic safety. Considered the mixed running characteristics of different types of vehicle moving on the urban road, Gaussian mixture model （GMM） was adopted to describe the vehicle speed distribution in free flow state. Taken the characteristics of bimodal distribution of field speed data into account, the mixed types of vehicle were divided into large vehicles and small vehicles. Then, GMM with two mixture components was presented, and the model parameters were estimated by the Expectation-Maximization （EM） algorithm. The results of calibration and verification with the measured data show that GMM could fit the observed speed data accurately and could distinguish the speeds of different types of vehicle effectively, which provided a theoretical basis for establishing speed limit of mixed vehicles and analyzing interaction between mixed vehicles.

关 键 词：交通工程 自由流速度 高斯混合模型 分布 

分 类 号：U491[交通运输工程—交通运输规划与管理]